Tuesday, May 22, 2012

Hepatitis B and C - Transient Elastography for Fibrosis Assessment Compared With Large Biopsies


Evaluation of Transient Elastography for Fibrosis Assessment Compared With Large Biopsies in Chronic Hepatitis B and C
 
Claudia Verveer; Pieter E. Zondervan; Fibo J. W. ten Kate; Bettina E. Hansen; Harry L. A. Janssen; Robert J. de Knegt
Posted: 05/22/2012; Liver International. 2012;32(4):622-628. © 2012 Blackwell Publishing

Abstract and Introduction
Abstract


Background Fibrosis determines prognosis and management in patients with chronic hepatitis B and C (CHB and CHC). Transient elastography (TE) is a promising non-invasive method to assess fibrosis. We prospectively studied the performance of TE compared to histology and also whether there are differences between CHB and CHC. Only large biopsies (≥25 mm) were used.

Methods We included 241 patients with CHB (n = 125) and CHC (n = 116), of whom we acquired 257 liver biopsies, all preceded by elastography. We correlated liver stiffness with fibrosis stage according to the METAVIR system, inflammation (Histology Activity Index), steatosis and iron. The impact of gender, age, body mass index, alcohol, alanine aminotransferase levels, platelet count, viral load and genotype on liver stiffness was evaluated.

Results The AUROC's for F ≥ 2 were 0.85 for CHB and 0.76 for CHC. AUROC's for F ≥ 3 were 0.91 for CHB and 0.87 for CHC and 0.90 and 0.91 for F4 for CHB and CHC respectively. For F ≥ 2 the cut-off value was 6.0 kPa for CHB and 5.0 kPa for CHC. The cut-off values for ≥F3 were 9.0 and 8.0 kPa for CHB and CHC, respectively, and 13.0 kPa for F4 in both CHB and CHC patients. Besides inflammation, all other remaining factors do not influence liver stiffness.

Conclusion For the diagnosis of fibrosis stages F ≤ 2 TE is suboptimal, and inflammation may induce higher values. For stages F ≥ 3 TE performance is good and equal in both CHB and CHC patients.

Introduction
The treatment of chronic viral hepatitis B (CHB) and hepatitis C (CHC) is directed towards improving life quality, survival and preventing the development of cirrhosis, complicated by hepatic failure and hepatocellular carcinoma (HCC).[1–6] Both conditions have a high mortality. As a result of many therapeutic options in both CHB and CHC, it is one of the greatest challenges to select the patients who benefit most from therapy. One of the parameters, which is used in the treatment criteria, is the stage of liver fibrosis, which can be estimated by histological analysis of the liver according to the fibrosis staging systems METAVIR,[7,8] Ishak[9] and others.[10,11] Currently, the presence of stage ≥F2 is a strong indicator for starting antiviral treatment[12–15]

Although the liver biopsy is regarded as the best standard for evaluation of fibrosis,[16] biopsy size and unequal distribution of fibrosis in the liver may influence its performance, resulting in an under- or overestimation of fibrosis stage with a possible negative impact on prognosis and indication for antiviral therapy. A biopsy with a smaller length has a larger sample size variability as well as intra- and interobserver variability.[17–19] Larger biopsies therefore increase the performance of fibrosis classification, but this is not feasible in clinical practice, for it may increase risk for complications. Studies suggest 25 mm as the optimal biopsy length for correct classification of the fibrosis stage.[18,20] It is well known that because of invasiveness and associated complications[4,21–24] liver biopsies are poorly tolerated by patients.

Transient elastography (TE) is a technique based on the combination of ultrasound and low-frequency waves for measurement of liver stiffness, as an indirect parameter for fibrosis, in a rapid and safe manner. The initial TE evaluation studies in CHC patients proved this technique to be an accurate tool for detecting severe fibrosis and cirrhosis.[25–29] However, there was considerable overlap present between liver stiffness measurements in patients with stages F1 and F2, resulting in a less discriminative power in the lower stages of fibrosis. Finally, the performance for the detection of stage ≥F2 was not optimal. These studies were limited by an imbalanced distribution of fibrosis, with an overrepresentation of stages F1 and F2, but only few F0. Also, the histology specimens were suboptimal and did not meet the criteria for optimal length.[18,20] Currently, there are only few publications on the performance of TE in CHB patients.[30–33]

The natural course of CHB is different from CHC and much more influenced by viral potency and immunological host factors. Where CHC has a more or less continuous progressive course, CHB often displays episodes of inflammation, characterized by a rise in ALT. Histological inflammation is likely to influence liver stiffness,[34,35] but also other factors contribute to liver stiffness.[36–38]
We evaluated the performance of TE in CHB and CHC patients by comparing histology, in large biopsies according to the advised international standards, and studied the impact of inflammation in a cohort of 257 paired measurements in 241 CHB and CHC patients. We evaluated the differences between the CHB and CHC patients regarding performance of TE. In addition, we tested the influence of patient and disease-related factors such as viral genotype, viral load, steatosis, iron, age, gender, weight, body mass index (BMI) and use of alcohol.

Patients and Methods
Patients
Patients with CHB or CHC for whom a liver biopsy was planned for the assessment of fibrosis and inflammation between January 2005 and May 2008 at the Erasmus University Medical Centre Rotterdam were asked to participate in this study. Patients were seen at the outpatient clinic for the liver biopsy and TE during the same session, in which TE preceded the liver biopsy. In 16 patients a second biopsy and a second elastography assessment were obtained at least 1 year later.
Inclusion criteria were the presence of CHB or CHC. Exclusion criteria were the coexistence of a second liver disease, co-infection with HIV, HCC and conditions known to cause a low success rate of TE such as ascites, obesity or narrow intercostal margins, which hamper proper placement of the TE probe. Antiviral therapy was not an exclusion criterion; however, in only six patients a biopsy was performed during antiviral therapy, these were all CHB patients, whereas in the remaining patients the biopsy was performed prior to therapy. The liver biopsy was considered adequately sized when the length was at least 25 mm. TE was performed with the FibroScan® (EchoSens, Paris, France). TE was considered reliable with eight or more successful acquisitions and a success rate ≥60%. In all other cases, TE was considered unreliable and measurements were excluded from analysis.
In all participating patients, serum bilirubin, albumin, alanine aminotransferase (ALT) and aspartate aminotransferase, gamma-glutamyltranspeptidase, prothrombin time and platelet count were determined prior to the liver biopsy. In addition, we tested the influence of patient or disease-related factors such as age, gender, weight, inflammation (by histological analysis and ALT determination), viral genotype, viral load, steatosis (according to the Brunt classification), iron, BMI and use of alcohol.

This study was approved by the Medical Ethical Committee of the Erasmus University Medical Center Rotterdam. All patients gave written informed consent.

Liver Biopsy and Histological Analysis
Two hepatologists, both with vast experience in liver biopsy procedures, performed the biopsies. Prior to the procedure, the liver was inspected with abdominal ultrasound for parenchymal and vascular abnormalities, focal lesions, vascular and biliary structures to reduce the risk of complications. The samples were taken with a true-cut needle (14 Ga; Angiotech, Gainesville, FL, USA). In some patients, two liver biopsies were required for obtaining an adequate sample size. There were no serious complications reported after liver biopsy in any of the patients. After biopsy, the sample was fixed in formalin, embedded in paraffin and cut lengthwise in 5-μm thick sections. The longest sections were stained with hematoxylin eosin (four sections) and picrosirius red (one section). All fixed specimens were inspected for adequate size (≥25 mm). These sections were used for fibrosis assessment with the METAVIR system (F0: no fibrosis, F1: portal fibrosis without septa, F2: few septa, F3: numerous septa without cirrhosis, F4: cirrhosis).[6,7] For the assessment of necro-inflammation the modified Histology Activity Index (HAI) was used[9] and for the determination of steatosis the Brunt classification.[39] Two expert liver pathologists scored the specimens for the different fibrosis categories as well as for the other histological features separately. In case of disagreement, a consensus was reached in a combined session.

Statistical Analysis
We assessed the predictive ability of TE in relation to histological fibrosis assessment with ordinal logistic regression analysis. The performance measures: sensitivity, specificity, positive predictive value, negative predictive value, the Youden Index and odds ratios[40] for a range of different cut-off values of TE in relation to histological fibrosis assessment were calculated. The cut-off values with the best discriminative ability between different fibrosis stages were determined by the maximum of the Youden Index, but with the criteria of a minimum of around 90% sensitivity of a true high fibrosis score. The stiffness data are presented on a logarithmic scale because of the skewed distribution.
The association of patient or disease-related factors on TE measurements were tested with regression analysis. Differences between groups were tested with Student's t-test. The analyses were performed with spss 16.0 (Statistical Package for Social Sciences, Chicago, IL, USA).

Results
Patient Characteristics
A total of 435 CHB and/or CHC patients were planned for a liver biopsy from January 2005 until May 2008. Two hundred and forty-one patients agreed and met the inclusion and exclusion criteria. In 50 cases, we did not obtain reliable TE measurements, and in 133 cases, we did not obtain adequately sized liver biopsies (≤25 mm), in 11 cases there were missing data. With 16 patients having a second biopsy and TE session, with an interval of at least 1 year, this finally resulted in a cohort of 257 paired measurements of TE, and biopsy in 241 patients with CHB or CHC.

The demographic and clinical patient characteristics are listed in Table 1. The distribution of CHB and CHC was 52 and 48%, respectively, with a male majority and predominantly Caucasian ethnicity in both groups. Except for an age difference there were no significant differences between the CHB and CHC groups regarding patient characteristics. Most patients had stage F1 or F2 fibrosis. As there were only few patients with stage F0, this group was combined with stage F1, indicated by stage F01.

Table 1. Characteristics of the study population
 
CharacteristicsCHB (n = 125)CHC (n = 116)P-value
Gender, M/F (%M)92/33 (73.6)77/39 (66.4)
Age (years)a36.5 (±13.0)45.4 (±9.5)<0.01
Race, n (%)
Caucasian61 (48.8)102 (88)
Asian41 (32.8)7 (6)
African23 (18.4)7 (6)
Weight (kg)a72.1 (±157)74.7 (±13.9)0.84
BMI (kg/m2)25.024.40.91
ALT (U/L)88.387.10.20
Platelet count (G/L)2052110.24
METAVIR stage of liver fibrosis, n (%)
F08 (5.7)7 (6)
F157 (40.7)38 (32.5)
F236 (25.7)41 (35)
F330 (21.4)22 (18.8)
F49 (6.4)9 (7.7)
Genotype hepatitis B
A27 (21.6%)
B13 (10.4%)
C25 (20.0%)
D40 (32.0%)
E9 (7.2%)
Not determined11 (8.8%)
HbeAg-positive49/67 (40.8%)
Genotype hepatitis C
174 (64.9%)
210 (8.8%)
320 (17.5%)
410 (8.8%)
Alcohol consumption
No73 (58.4%)49 (42.2%)
<2 glasses a week35 (28.0%)29 (25.0%)
1–4 glasses a day7 (5.6%)10 (8.6%)
>4 glasses a day2 (1.6%)9 (7.8%)
Previous abuse8 (6.4%)19 (16.4%)

*Mean value (±1 SD).
n, number of liver samples and accompanying TE measurements; ALT, alanine aminotransferase; BMI, body mass index; CHB, chronic viral hepatitis B; CHC, chronic viral hepatitis C.

Stiffness Compared to Histological Fibrosis
The box plot in Fig. 1 represents the stiffness values (kPa) for the various stages of fibrosis in CHB and CHC, according to the METAVIR system. The results of the ordinal logistic regression of all stages of fibrosis 0–4, in patients with CHB and C as well as the cumulative prediction of fibrosis score by kPa in both patient groups are presented in Fig. 2. Although the Youden index was optimal at 7.0 kPa for both CHB and CHC, there was also a substantial misclassification of patients (14.67% in CHB and 40.28% in CHC patients). To obtain sensitivity >90%, we had to choose a cut-off value of 6.0 kPa in CHB and 5.0 kPa in CHC. Similarly, the cut-off values for ≥F3 were 9.0 and 8.0 kPa for CHB and CHC, respectively, and 13.0 kPa for F4 in both CHB and CHC patients.




Figure 1.
Liver stiffness for each METAVIR fibrosis stage. The box indicates the lower and upper quartiles and the horizontal line is the median (50th) percentile. The distance between the top and the bottom of the box represents the interquartile range (IQR). The points at the ends of the 'whiskers' are the 2.5 and 97.5% values.


Figure 2.
(a) This figure presents the results of the ordinal logistic regression of all stages of fibrosis 0–4, in patients with chronic viral hepatitis B (CHB). The prediction of fibrosis score: stages F0, F1, F2, F3 and F4 by kPa-values in patients with chronic viral hepatitis C (CHC). For example if kPa = 8, the prediction of stage F0 is 2%, F1 is 42%, F2 is 38%, F3 is 16% and F4 is 2%, respectively, and F0 or F4 about 0%. Note the sum of the predictions for a fixed kPa value = 100%. (b) This figure presents the cumulative prediction of fibrosis score by kPa in patients with CHB. The cumalative prediction of fibrosis score F01, F234, F34, and F4 by kPa in patients with CHB. For example if kPa = 8 the prediction of a F234 is almost equal to a F01. (c) This figure presents the results of the ordinal logistic regression of all stages of fibrosis 0–4 in patients with CHC. The prediciton of fibrosis score: stages F0, F1, F2, F3 and F4 by kPa-values in patients with CHC. For example if kPa = 8, the prediction of a stage F0 is 2%, F1 28%, F2 48%, F3 20% and F4 2%. Note the sum of the predictions for a fixed kPa value = 100%. (d) This figure presents the cumulative prediction of fibrosis score by kPa in patients with CHC. The cumulative prediction of fibrosis score F01, F234, F34 and F4 by kPa in patients with CHC. For example if kPa = 8, the prediction of a F234 is 72%, much higher than a F01 which is 26%.

The AUROC was 0.85 in CHB patients for F ≥ 2 and 0.76 for CHC. AUROC's for F ≥ 3 were 0.91 for CHB and 0.87 for CHC, and 0.90 and 0.91 for F4 in CHB and CHC respectively.

Influence of Inflammation and Steatosis on Transient Elastography Performance
As studied with HAI for chronic viral hepatitis, biopsies of CHB patients showed significantly less inflammation than those of CHC patients (P < 0.001), with a median HAI of 5 (minimum 1, maximum 10) for CHB and 6 (minimum 1, maximum 11) for CHC. Patients with F ≥ 2 had more extensive inflammation in both patient groups compared with those with F0 or F1. The histopathological findings concerning biopsy length, inflammation and steatosis for both groups are listed in Table 2. The mean lengths of the biopsies and the grade of steatosis, classified by Brunt, were not statistically significantly different between CHB and CHC patients, or between fibrosis stages. However, inflammation increased liver stiffness regardless of fibrosis stage (P < 0.001).


Table 2. Histology findings in the study population
 
METAVIRLength of biopsy (mm)aNecro-inflammatory index (HAI, Ishak)aSteatosis (Brunt)a
Stage
F0 (n=15)
CHB (n=8; 3.1%)30.0 (4.1)3.12 (1.25)0.50 (0.55)
cCHC (n=7; 2.7%)34.0 (4.3)3.86 (1.68)0.43 (0.54)
F1 (n=95)
CHB (n=57; 22.2%)30.7 (4.9)4.40 (1.58)0.77 (0.76)
CHC (n=38; 14.8%)31.0 (5.1)5.74 (1.72)0.71 (0.77)
F2 (n=77)
CHB (n=36; 14%)30.2 (3.3)5.50 (1.63)0.69 (0.92)
CHC (n=41; 16%)31.6 (4.7)6.10 (1.58)0.73 (0.74)
F3 (n=52)
CHB (n=30; 11.7%)30.4 (4.8)6.40 (1.94)0.73 (0.69
CHC (n=22; 8.6%)31.1 (5.1)6.55 (2.15)0.77 (0.61)
F4 (n=18)
CHB (n=9; 3.5%)30.1 (3.9)7.33 (2.00)0.67 (0.50)
CHC (n=9; 3.5%)30.2 (5.9)6.56 (1.01)0.78 (0.97)
*Mean value (±1 SD).
CHB, chronic viral hepatitis B; CHC, chronic viral hepatitis C; HAI, Histology Activity Index.


Although not as strong as HAI, increased serum ALT values were also associated with increased liver stiffness (P = 0.002). Other patient and disease-related factors, such as gender, age, BMI, alcohol, platelet count, viral load and genotype were not associated with liver stiffness.
Steatosis influenced liver stiffness; this was however without statistical significance (P = 0.59). All other previously mentioned investigated patient and disease-related factors did not influence liver stiffness.

Discussion
We compared TE with histology, according to the METAVIR system, in 257 paired measurements from 241 patients with CHB or CHC, using biopsies of at least 25 mm, to assess performance of TE for fibrosis assessment. In this cohort, we investigated differences in liver stiffness between CHB and CHC, and studied the patient- and disease-related factors associated with liver stiffness. This is the first independent evaluation study, including large biopsies combined with paired TE measurements in patients with CHB and CHC, allowing improved evaluation of TE. Our study also shows a good to excellent performance of TE for F ≥ 3 and F4. These results are comparable to the results found in previous studies. However, the results concerning F ≥ 2 (0.85 in CHB patients and 0.76 for CHC) should be considered suboptimal.[41] Our cut-off values are slightly different as compared with previous TE studies,[25,27,28] mainly concerning CHC, with 5.0 kPa for F ≥ 2 and 8.0 kPa for F ≥ 3. This was caused by the misclassification in TE of a substantial number of patients with stage F2 as F1, due to more severe inflammation in CHC patients. TE predicted stages F3 and F4 with equal performance in patients with CHB and CHC. Bedossa et al.[20] showed that with large biopsies, fibrosis is staged accurately in 75% of all samples, therefore part of the misclassification might be due to the limitations of both the liver biopsy as well as the METAVIR system. In contrast to our expectation, the use of large liver biopsies did not improve performance of elastography compared with the literature.

We found higher liver stiffness in CHB patients than in CHC patients, although not statistically significant, especially for stages F01 and F2 (see Fig. 1.) Several earlier studies in CHB patients also exhibited different liver stiffness values as opposed to CHC patients, but they were usually lower.[31,42,43] Differences in liver stiffness values are likely to arise because of differences in specific inflammation patterns in CHB and CHC, as seen in histopathology. However, our CHB patients had less severe inflammation than our CHC patients therefore we are not able to explain the differences in liver stiffness between both groups.

We also demonstrated that inflammation increases liver stiffness, causing overestimation of the real fibrosis stage. In this cohort, CHB patients had less histological and biochemical signs of inflammation than CHC patients (P < 0.001). Steatosis according to Brunt did not influence liver stiffness. However, the Brunt classification is not well adapted to viral hepatitis and this poses a limitation to our study, especially concerning the statistical analysis.

Moreover, in the first TE studies the influence of inflammation was not investigated, but reported in later studies.[31,34,35,38,44–47] In all METAVIR stages, liver stiffness is significantly influenced by inflammation as well as by serum ALT level. In the lower stages of fibrosis F1 and F2, inflammation seems to be an important factor associated with increasing liver stiffness, therefore causing a decrease in performance of fibrosis assessment with TE, and considerable overlap between stages F1 and F2. Also in these studies, and in our study, there was an imbalanced distribution of fibrosis, with an overrepresentation of stages F1 and F2, but only few F0 patients. A limitation was the use of liver biopsies with inadequate length, according to international standards.

According to our study results, TE can be used in clinical practice for the evaluation of fibrosis and cirrhosis. When liver stiffness equals or exceeds 14 kPa, this can be regarded as confirmative for cirrhosis, in both CHB and CHC patients. Severe fibrosis is with high probability present when liver stiffness is between 9 and 14 kPa in CHB patients, and between 8 and 14 kPa for CHC patients. However, for the diagnosis of stage F ≤ 2, the performance of TE is suboptimal and less reliable, and inflammation increases liver stiffness. In all patients the determination of ALT levels, as an indirect parameter for inflammation, is therefore especially important because inflammation influences liver stiffness. We conclude that TE is an accurate tool to assess severe fibrosis (stages F3 and F4) in CHB and CHC patients.

http://www.medscape.com/viewarticle/762116

No comments:

Post a Comment