1259/bjr/75208448
Ultrasound evaluation of the brosis stage in chronic liver disease by the simultaneous use of low and high frequency probes
1
T NISHIURA, RMS, 1H WATANABE, RMS, 1,2M ITO, MD, 3Y MATSUOKA, MD, 4K YANO, MD, 4 M DAIKOKU, MD, 4H YATSUHASHI, MD, 4K DOHMEN, MD, FACP, SJSUM and 4 H ISHIBASHI, MD, FACP, SJSUM
1
Clinical Laboratory, 2Department of Pathology, 3Deartment of Radiology and 4Clinical Research Centre, NHO National Nagasaki Medical Centre, Omura, Nagasaki, 856-8562 Japan
Abstract. A liver biopsy is currently considered the denitive diagnostic modality for establishing the severity of hepatic brosis. We analysed the diagnostic sensitivity and accuracy of ultrasound (US) using both low frequency and high frequency probes as a repeatable, inexpensive, and reliable method to determine the brosis stage in chronic liver disease and then compared our results with the histological ndings. A total of 103 patients with chronic liver disease (60 males and 43 females, average age 51 years old) who had undergone both a liver biopsy and US with 25 MHz frequency and 512 MHz frequency probes were prospectively evaluated in this study. An US scoring system using both the low frequency and high frequency probes was performed by evaluating the edge, surface and parenchymal texture of the liver. Each score was obtained by evaluating three parameters; the bluntness of the liver edge, the irregularity of the surface and the coarseness of the parenchymal texture were evaluated and then compared with the histological ndings. The US scores of the liver edge (rs: 0.6668), liver surface (rs: 0.9007) and liver parenchymal texture (rs: 0.8853) correlated signicantly with the brosis stage obtained based on the biopsy ndings. The accumulated US scores of these three parameters, however, was found to be the most reliable indicator (rs: 0.9524). Patients with an accumulated score of 6.5 or more were all found to have brosis stage 4 in which the accuracy of our scoring system for correctly predicting cirrhosis was found to be 100% sensitive. When an accumulated US score of 3 was interpreted to indicate mild brosis (a brosis score of 0 or 1), all 42 patients with stage 0 or 1 brosis were found to have an accumulated US score of 3 or less (a probability of 100%) and 42 of 53 patients with a score of 3 or less were found to have stage 0 or 1 brosis (specicity of 79.2%). An ultrasound evaluation of the liver brosis stage based on the scoring system using both low and high frequency probes was found to be a reliable and effective alternative to the histological staging in chronic liver diseases.
The liver brosis stage in patients with chronic liver diseases due to an infection with hepatitis B virus (HBV) or C virus (HCV) is a pivotal factor regarding both the therapeutic options and for predicting the prognosis. A liver biopsy is considered to be the gold standard for diagnosing the liver brosis stage and predicting the outcome of the diseases. Although a percutaneous liver biopsy is relatively safe, it is still associated with a risk of complications, patient discomfort and a high cost. In addition, liver biopsy examinations may lead to false negative results due to inadequate liver tissue sampling. Therefore, there is a need to develop a simple, reliable and non-invasive modality in order to assess the liver brosis stage [1]. Ultrasound (US) is a non-invasive, inexpensive and repeatable modality and has been used as the most important and valuable diagnostic tool for detecting hepatocellular carcinoma (HCC) during the follow-up of patients with viral hepatitis [2, 3]. US is also used for monitoring the response of HCC to treatment.
Received 22 July 2003 and in nal form 31 August 2004, accepted 5 October 2004. Address correspondence to Dr Koji Yano, Clinical Research Centre, NHO National Nagasaki Medical Centre, Kubara 2-1001-1 Omura, Nagasaki 856-8562 Japan.
An ultrasound evaluation of the liver brosis stage of chronic liver disease has been performed by assessing various ultrasound factors such as the liver size, the bluntness of the liver edge, the coarseness of the liver parenchyma, nodularity of the liver surface, the size of the lymph nodes around the hepatic artery, the irregularity and narrowness of the inferior vena cava, portal vein velocity or spleen size [1, 48]. However, the conventional denition of the brosis stage of the liver based on evaluation of these ultrasound factors is imperfect and lacks accuracy and reliability. Furthermore, these ndings also depend largely on the equipment used [8]. Indeed, a few reports have demonstrated no consistent correlation between the grey scale ultrasound ndings and the histological ndings, thus claiming that grey scale US is unreliable for grading and staging the degree of liver damage [9]. However, recent advances in US technology have improved the diagnostic accuracy for brosis in patients with chronic liver disease. Therefore, we carried out a study to evaluate the accuracy of the liver brosis stage by utilizing the techniques of advanced ultrasound performance in 103 patients with chronic liver disease and compared the results obtained with the histological ndings.
189
score of 2 was given for a blunted edge and a score 3 for severe irregular surface or a highly coarse texture when these characteristics were clearly conrmed by the low frequency probe. In cases in which three parameters such as the edge, surface and parenchymal texture could not be determined to be either mild or severe using a low frequency probe, then the high frequency probe was used to determine whether they were mild or severe. The high frequency probe was used to obtain a score 0 or 1 because the sensitivity obtained by the high frequency probe was superior to the one by the low frequency probe regarding mildly abnormal changes such as a score 0 or 1. Conversely, regarding such advanced changes as a score of 2 or 3, the low frequency probe was more useful because the probability obtained by the low frequency probe was superior to that by the high frequency probe. As a result, the brosis stage predicted from the accumulated scores of the liver edge, surface and parenchymal texture by US were considerably more reliable than the individual scores of these three parameters when they were compared with the histological ndings.
US system
The patients were studied ultrasonically using a realtime apparatus (HDI 5000 Sono CT, ATL, USA) with a 25 MHz convex array transducer C52 (low frequency probe) and a 512 MHz convex array transducer L125 (high frequency probe).
Histological ndings
Liver biopsy specimens were obtained from the anterior segment of the right lobe in each patient, using a 16-gauge Sonopsy-C1 biopsy needle (Hakko Co., Tokyo, Japan). All of the histological slides were reviewed by an experienced pathologist without any knowledge of the clinical details or the US ndings. The New Inuyama scoring system for chronic hepatitis was proposed by the Japanese Liver Study Group in 1994 [11], which is similar to the classication of chronic hepatitis determined by the USA-European Liver Study Group [12]. The New Inuyama scoring system was used to assess the brosis stage as follows; score 0: no brosis, score 1: brous portal expansion, score 2: bridging brosis, score 3: bridging brosis with lobular degeneration, and score 4: cirrhosis.
Statistics
The Spearmans correlation test was used to assess any correlations between the liver brosis stage and the US scoring system. The sensitivity, specicity and positive predictive values of the US scoring system were calculated and compared with the results of the liver brosis stages.
Results
US scoring system and brosis stage The relationship between the liver edge and brosis (Figure 4) Of 11 patients with a sharp liver edge (edge score 0), 6 patients (55%) were found to have stage 0 brosis and 5 (45%) in stage 1 brosis. The liver edge score of 34 patients
Table 1. Findings for the ultrasound features of the edge, surface and parenchymal texture of the liver
Score 0 Edge Surface Parenchymal texture sharp smooth ne Score 1 mildly blunted edge mildly irregular mildly coarse Score 2 blunted edge irregular coarse Score 3 highly irregular highly coarse
190
(a)
(b)
(c)
edge with a high frequency probe, (b) a mildly blunted edge with a high frequency probe, and (c) a blunted edge with a low frequency probe.
in an early brosis stage (stage 1) showed various results; 5 patients had a score of 0, 6 a score 0.5, 9 a score of 1, 5 a score of 1.5 and 9 a score of 2. On the other hand, the patients with stage 2, 3 and 4 brosis correlated well with the edge score and were categorized into a blunted edge (score 2); 26 of 26 patients with stage 2 brosis, 12 of 13 patients with stage 3 brosis, 22 of 22 patients with stage 4 brosis. The liver edge and the brosis score thus showed a statistically signicant correlation (rs: 0.6668).
score of 1 or less. 26 patients with the brosis stage 2 had different results regarding the surface score, i.e. 3 patients had a surface score of 0, 7 a score of 0.5, 9 a score of 1, 4 a score of 1.5 and 3 a score of 2, respectively. 12 of 13 patients with stage 3 brosis were found to have a surface score of either 1.5 or 2. All 22 patients with the stage 4 brosis were found to have a surface score of 2 or more. The liver surface score and the brosis stage showed a statistically signicant correlation (rs: 0.9007).
The relationship between the liver surface and brosis (Figure 5) Eight patients with stage 0 brosis were all put into the smooth liver surface group (surface score 0), and 34 patients with the brosis stage 1 were all found to have a
The British Journal of Radiology, March 2005
The relationship between liver parenchymal texture and brosis (Figure 6) Eight patients with stage 0 brosis were all found to have a ne liver parenchymal texture (parenchymal score 0), and 29 of 34 patients with stage 1 brosis were found
191
(a)
(b)
(c)
(d)
Figure 2. The ultrasound features of the liver surface; (a) a smooth surface with a high frequency probe, (b) a mildly irregular surface with a high frequency probe, (c) an irregular surface with a low frequency probe, and (d) a highly irregular surface with a low frequency probe.
to have a score of 0. Of 26 patients with stage 2 brosis, 11 had a score of 0, 14 a score of 1 and 1 a score of 2. Five and 6 patients of the 13 patients with stage 3 brosis had a score of 1 and 2, respectively. All 22 patients with stage 4 brosis had a parenchymal score of 2 or more. The liver parenchymal score and the brosis stage showed a statistically signicant correlation (rs: 0.8853).
Relationship between the brosis grade and the accumulated scores of the liver edge, surface and parenchymal texture (Figure 7) The liver edge, surface and parenchymal texture scores were all determined and compared with the histological
192
brosis score. Of eight patients with brosis stage 0, six patients were found to have a total score of 0 and while two had a total score of 1. 34 patients with stage 1 brosis had a total score 1.53 consisting of 2 patients with scores of 0, 11 and a score of 0.51, 17 had a score of 1.52.0, and 4 had a score of 2.53, respectively. 26 patients with stage 2 brosis shifted to the progressive brosis stage, such as 1 patient with a score of 1.52, 10 with a score of 2.53, 9 with a score of 3.54, and 6 with a score of 4.55, respectively. Similarly, 13 patients with stage 3 brosis shifted to a progressive brosis stage such as 1 patient with a score of 3.54, 5 with a score of 4.55, and 7 with a score of 5.56, respectively. Out of 22 patients with stage 4 brosis had a score of 6.57.0 and eleven had a score of
The British Journal of Radiology, March 2005
(a)
(b)
(c)
(d)
Figure 3. Scores for the ultrasound features of the liver parenchymal texture; (a) ne parenchymal texture with a high frequency
probe, (b) a mildly coarse parenchymal texture with a high frequency probe, (c) a coarse parenchymal texture with a low frequency probe, and (d) a highly coarse parenchymal texture with a low frequency probe.
77.5, respectively. The brosis grade and the accumulated scores of these three parameters were more signicantly correlated (positive predictive value, 0.9524) than the correlations of each score alone.
Discussion
Chronic liver diseases with viral infection manifest varying degrees of hepatic brosis ranging from no brosis to cirrhosis. Yoshida et al revealed that the annual
The British Journal of Radiology, March 2005
incidence of hepatocellular carcinoma increased from 0.5% among patients with the stage F0 or F1 brosis to 7.9% among the patients with stage F4 brosis [13]. It has thus become increasingly apparent that the brosis stage is a key factor in dening the prognosis and management of chronic liver diseases with a viral infection. The gold standard in hepatology for the diagnosis of the brosis stage has been a histological liver evaluation based on specimens taken either by a needle biopsy or at operation. Recently, non-invasive and reliable assessments for monitoring chronic liver disease using the platelet
193
Figure 5. Relationship between the liver surface and brosis ndings. 194 The British Journal of Radiology, March 2005
Figure 6. Relationship between the liver parenchymal texture and brosis ndings.
Figure 7. Relationship between the accumulated ultrasound score and brosis score ndings. The British Journal of Radiology, March 2005 195
counts [1416], aspartate aminotransferase (AST)/alanine aminotransferase (ALT) ratio [15, 16], and serum hyaluronan and type III procollagen amino-terminal peptide [17] have been developed. However, none of the currently available tests or modalities can completely replace a histological analysis. Previous studies have assessed several methods for evaluating the brosis stage of chronic liver disease using various US parameters. However, there have so far been few studies concerning the accuracy in detecting the signs of compensated cirrhosis by US [5, 18]. Gaiani et al [5] and Hung et al [19] proposed a complex US scoring system using indices of the liver surface, parenchymal echogenecity, the vessel pattern, spleen size etc. to determine the brosis stage. In addition, recent advances in ultrasound technology have now made it possible to obtain more precise information about the liver surface, edge and parenchymal texture [8]. Therefore, we conducted this study to clarify whether the US scoring system with a newly developed US equipment based on the conventional parameters of the liver edge, surface and parenchymal texture might obtain sufciently accurate results in comparison with the histological ndings for brosis obtained by a liver biopsy. In this prospective study, among these parameters such as the liver edge, liver surface and liver parenchymal texture, the liver edge was not as specic for evaluating liver brosis as the liver surface and parenchymal texture in our study because a mildly blunted (score 1) or blunted edge (score 2) was frequently found in the early brosis stage (stage 1) (67.6%). On the other hand, the liver surface and liver parenchymal texture obtained by US showed a better correlation with the histological ndings (correlation coefcient of 0.9007 in the liver surface, and 0.8853 in the parenchymal texture). With conventional US, the liver surface has been most commonly utilized as a sole indicator for the diagnosis of cirrhosis [5, 2022]. However, numerous papers have reported that the sole factor of the liver surface can not sufciently distinguish cirrhosis from chronic hepatitis. Gaiani et al conrmed that the stage of cirrhosis may be underestimated when based on a single specimen and claried that only two US variables, namely liver surface nodularity and the portal vein mean ow velocity, independently contributed to the diagnosis of cirrhosis [5]. In our study, all seven patients with a highly irregular surface were found to have cirrhosis (stage 4 brosis) histologically. On the other hand, 15 of 22 patients (68.2%) with cirrhosis were found to have an irregular surface (score 22.5), not a highly irregular surface (score 3). Indeed, the results of our study showed a signicant correlation between the ultrasound liver surface and the histological brosis stage. An irregular and nodular liver surface may be easily assessed in patients with decompensated liver cirrhosis, particularly in the case of ascites, and it has been observed in 88% of unselected patients with cirrhosis [20]. Gaiani et al reported the ndings of a US scoring system, based on the liver, spleen and portal vein features, which identied cirrhosis in 82.2% of the cases [5]. In our study, both the right and the left liver lobes were evaluated for each factor, and the average score for each parameter was calculated. Both the low frequency and the high frequency probes were used to limit the subjective nature of the assessment of the US ndings of various degrees. In
196
this prospective study, US was performed with the simultaneous use of low frequency and high frequency probes to determine the sensitivity and probability according to the characteristics of ultrasound. Although our study was limited on account of the relatively small number of patients due to the strict inclusion criteria, 22 patients with an accumulated score of 6.5 or more were all found to have a brosis score of 4. Therefore, our scoring system for correctly predicting cirrhosis was found to be 100% sensitive. Furthermore, although the major drawback with US in comparison with the liver histology has been considered to be the failure to detect mild brosis or none at all, our scoring system thus provided relatively accurate information about liver brosis. When the accumulated US score of 3 was interpreted as mild brosis (a brosis score of 0 or 1), all 42 patients with stage 0 or 1 brosis were categorized into an accumulated US score of 3 or less (a probability of 100%) and 42 of 53 patients with a score of 3 or less were found to have stage 0 or 1 brosis (specicity of 79.2%). In addition, the score proposed in our study is easy to obtain and can be applied in every ultrasound laboratory by utilizing regular commercially available US equipment. Evaluating the ultrasound pattern using either one or two parameters becomes much more complex at the stage of chronic liver disease than that of complete cirrhosis. Our scoring system based on three parameters such as the liver edge, surface and parenchymal texture was able to accurately predict the brosis stage (correlation coefcient of 0.9524), especially when distinguishing chronic hepatitis from compensated liver cirrhosis. When an exclusion of liver cirrhosis is requested, then US alone is therefore considered to provide sufcient information based on this scoring system. Furthermore, if a histological analysis can not determine the brosis stage correctly due to fragmentation or architectural distortion, then this ultrasound diagnostic modality of brosis could replace a histological diagnosis. In conclusion, we demonstrated that our US scoring system is clinically useful for differentiating patients who have chronic liver disease with minimal or no brosis from those with mild to severe brosis. These parameters may also be useful for providing prognostic information and also for determining the optimal therapeutic options during the follow-up of patients with chronic liver disease, especially in patients with chronic hepatitis C or B, in order to predict the occurrence of HCC. In addition, further study is called for to determine whether or not the wider use of this scoring system could apply to other forms of hepatic brosis such as those suffering from long-term hepatotoxic disease, congenital diseases in children and non-viral infective forms of chronic liver disease in order to obtain an improved response to therapy.
References
1. Fontana RJ, Lok ASF. Noninvasive monitoring of patients with chronic hepatitis C. Hepatology 2002;36:S57S64. 2. Dohmen K, Shirahama M, Onohara S, Miyamoto Y, Torii Y, Irie K, et al. Differences in survival based on the type of follow-up for the detection of hepatocellular carcinoma: an analysis of 547 patients. Hepatol Res 2000;18:11021.
197