[English only, Educational] Assessment of fibrosis in chronic liver diseases

StephenW

http://onlinelibrary.wiley.com/doi/10.1111/j.1751-2980.2008.00356.x/full?globalMessage=0
Assessment of fibrosis in chronic liver diseases

• Kun ZHOU,
• Lun Gen LU
  

Keywords:

• liver biopsy;
• liver fibrosis;
• non-invasive test;
• serum marker panel;
• transient elastography
  

Abstract
The assessment of liver fibrosis provides useful information not only for diagnosis but also for therapeutic decisions. Although liver biopsy is the current gold standard for fibrosis assessment, it has some risks and limitations, including intra-observer and inter-observer variation, sampling error and variability. In recent years, many studies and great interest have been dedicated to the development of non-invasive tests to substitute a liver biopsy for fibrosis assessment and follow up. Advances in serological and radiological tests such as serum marker panels, transient elastography and their combinations can assess fibrosis accurately and reduce the need for a liver biopsy. But at present, all have failed to completely replace a liver biopsy because of their respective limitations and an imperfect gold standard used in current researches. The searching for an ideal surrogate is still in progress.


INTRODUCTION
Chronic liver diseases (CLD) are very common. More than 400 million individuals worldwide are chronically infected with hepatitis B virus (HBV), and more than 170 million are chronically infected with hepatitis C virus (HCV). Nonalcoholic fatty liver disease also attracts more and more attention, paralleling the increasing prevalence of obesity, diabetes mellitus and the metabolic syndrome, and alcoholic liver disease (ALD) is prevalent in drinkers. Advanced fibrosis and cirrhosis develop in about 20–40% of patients with CLD, some of whom finally progress to end-stage liver disease or hepatocellular carcinoma. But the progression may take years or decades. A CLD natural history is variable and its long-term evolution differs in individual patients.

Previously, fibrosis was thought to be an irreversible process and received little attention until the 1980s, when it was discovered that hepatic stellate cells (HSC) played an important role during fibrosis. Following chronic liver injury in CLD, HSC proliferate and activate under inflammatory milieu, acquiring proinflammatory and fibrogenic properties. Activated HSC synthesize large amounts of extracellular matrix (ECM) constituents. The accumulation of ECM distorts the hepatic architecture by forming a fibrous scar, and the subsequent development of nodules of regenerating hepatocytes defines cirrhosis. In the 1990s, it was demonstrated that liver fibrosis may be reversible when the cause is treated. Since, researchers have been stimulated to develop therapies for treating hepatic fibrosis. Although successful treatments, such as antiviral therapies for HBV and HCV, seems to modify fibrosis and prevent progression to cirrhosis and cancer, their response rates are low (especially in HCV genotype 1 patients) and with frequent side effects and high costs. Thus, assessing fibrosis precisely is the main determinant in the management of CLD, to decide when to begin a treatment and assess treatment response. Up to now we have developed three main tools to assess fibrosis: liver biopsies, serological and radiological tests. Unfortunately, they all have limitations and pitfalls. To discuss their advantages and deficiencies will be helpful in scientific research and clinical practice.


INVASIVE MEASUREMENTS
Liver biopsyThe liver biopsy has been considered the gold standard for confirming a clinical diagnosis, for assessing the severity of necro-inflammation and fibrosis, for identifying cofactors and comorbidities, and for monitoring the efficacy of treatment ever since the first liver biopsy was performed by Paul Ehrlich in 1883.1 The procedure is particularly useful for diagnosing the early stages of fibrosis and identifying patients at high risk of progressing to fibrosis, but it has also a number of limitations. Patient acceptance is somewhat low because a biopsy is expensive, invasive and associated with some discomfort and complications. Pain appears in about one-quarter of patients, and other complications include bleeding, biliary peritonitis, pneumothorax and a mortality rate of about 0.01%.2 A sampling error of at least 24% is reported, usually because of specimen fragmentation or inadequate length. Colloredo et al. concluded that an optimum specimen should be at least 20 mm in length with 11 complete portal tracts.3 Even with adequate-sized biopsies, the interpretation might be unreliable, because the distribution of necro-inflammation and fibrosis is not homogeneous and a liver biopsy samples only 0.00002 of the mass of the liver.

Several semi-quantitative scoring systems have been proposed to describe and quantify necro-inflammation, steatosis and fibrosis in the liver, particularly for chronic viral hepatitis. These include the Knodell histological activity index (HAI) first proposed in 1981, then modified to the Scheuer system, the METAVIR system and the Ishak modified HAI.4 However, all these scoring systems could only provide qualitative descriptors to stage fibrosis, and the staging of certain histopathological changes differ in different systems (Table 1). This could cause considerable intra-observer and inter-observer variation and difficulty in comparison.

Table 1.                 Scoring systems for staging fibrosis

Pathologic features	Knodell	Scheuer	METAVIR	Ishak
No fibrosis	0	0	0	0
Enlargement of some portal tracts	1	1	1	1
Enlargement of most portal tracts	1	1	1	2
Periportal septa	1	2	1	2
Occasional portal– portal septa	3	2	2	3
Numerous septa (portal–portal and/ or portal–central)	3	3	3	4
Occasional nodules	4	4	4	5
Definite cirrhosis	4	4	4	6

Using computerized digital image analysis, the amount of fibrosis in liver biopsy specimens can be evaluated by a quantitative score. Though this is thought to be less reliable in determining early stage fibrosis, recent advances such as a higher resolution digital camera can improve discrimination between the varying stages of liver fibrosis, including mild fibrosis. It may be a more precise method than semi-quantitative histological stages for monitoring fibrosis progression or regression during clinical therapeutic trials.5 Considering the irregular shape of specimens, fractal and spectral dimension analysis can also be used to improve accuracy.6

The detection of genes correlated with fibrosis from biopsy samples has renewed interest in liver biopsies. Changes in liver gene expression can indicate fibrosis progression precisely at an early stage.7 Genetic studies have identified possible genetic polymorphisms that influence the progression of liver fibrosis.8 The identification of panels of key genes correlating with differences in the progression of CLD could lead to establishing excellent prognostic/diagnostic tools.


Hepatic venous pressure gradient (HVPG)HVPG, as an expression of intrahepatic resistance, does not exceed 5 mmHg in the absence of significant fibrotic evolution. The measurement of HVPG is a validated, safe and highly reproducible technique. It may be considered a dynamic marker of disease progression in patients with HCV and an end point in antiviral therapy, irrespective of the antiviral response.9 However, the technique is invasive and expensive; it requires technical expertise and has a low patient acceptance.



SEROLOGICAL TESTS
The limitations of liver biopsy led to the search for non-invasive tests to assess liver fibrosis. Afdhal and Nunes et al.10 suggest the following criteria for an ideal marker of liver fibrosis: it should be liver specific; it should not be influenced by alterations in liver, renal, or reticulo-endothelial function; it should measure one or more of the processes related to fibrosis (the stage of fibrosis, the activity of matrix deposition, or the activity of matrix removal) and should be easy to perform.

Direct serum markersThe key step in the pathophysiology of liver fibrosis is a balance between ECM deposition and removal. The accumulation of ECM results from both increased synthesis and decreased degradation. The principal ECM constituents are synthesized by activated HSC, while they are broken down by a family of enzymes known as matrix metalloproteinases (MMP). Many studies have been dedicated to find serum ECM markers for fibrosis assessment: (i) collagens: N-terminal peptide of type pro-collagen (PIIINP), type IV collagen 7s domain(IV-7S); (ii) proteoglycans: hyaluronic acid; (iii) glycoproteins: laminin, human cartilage glycoprotein 39; (iv) collagenases and their inhibitors: MMP, tissue inhibitor of metalloproteinases; and (v) cytokines: transforming growth factor β, platelet-derived growth factor (PDGF), tumor necrosis factor β.

The clinical applications of such markers appear innovative and they are useful for assessing the speed of liver fibrogenesis and estimating the response to antiviral therapies or anti-fibrotic drugs. But most of them are insensitive in milder fibrosis, and it must be stressed that these markers reflect fibrogenesis and fibrolysis more than fibrosis itself. In other words, there may be a highly active fibrotic process in the liver, although fibrotic tissue has not yet been developed, or there may be heavy fibrosis in the liver but fibrotic activity is temporarily discontinued.

StephenW

Serum marker panelsSince present direct markers cannot satisfy yet the clinical need of measuring the fibrosis, an alternative approach turn out to be combining a number of serum markers to generate algorithms capable of evaluating fibrosis. A large number of panels have been studied by groups worldwide11–32 (Table 2).

Table 2.                 Studies of serum markers panels for assessment of liver fibrosis

Index, author, year, reference	Patients (n)	CLD	Markers in panel	AUROC (T-V)†
†                         The area under the receiver operating characteristic curves (AUROC) for the diagnosis of significant fibrosis (stage 2–4 by the METAVIR or Scheuer classification, 3–6 by the Ishak score). T-V, AUROC values of training group and validation group. ‡                          ‡Differentiation cirrhosis from no cirrhosis. §                         Differentiation advanced fibrosis (Ishak 4–6) from mild to moderate fibrosis (Ishak 0–3). ¶                         Differentiation advanced hepatic fibrosis (defined as F3–F4 by METAVIR) from milder (F0–F2). A2M, α2-macroglobulin; AAR, AST/ALT ratio; ALP, alkaline phosphatase; ALT, alanine aminotransferase; apoA1, apolipoprotein A1; APRI, aspartate aminotransferase to platelet ration index; AST, aspartate aminotransferase; AUROC, area under the receiver operating characteristic curves; BMI, body mass index; CLD, chronic liver disease; GGT, γ-glutamyltransferase; HA, hyaluronic acid; HBV, hepatitis B virus; HCV, hepatitis C virus; NAFLD, Nonalcoholic fatty liver disease; HIV, human immunodeficiency virus; HOMA-IR, homeostasis model assessment insulin resistance (fast glucose × plasma gluc/22.5); Hpt, haptoglobin; INR, international normalized ratio; IV-7S, type IV collagen 7s domain; MMP-1, metalloproteinase 1; Mn-SOD, manganese superoxide dismutase; PCHE, pseudocholinesterase; PI, prothrombin index; PIIINP, N-terminal peptide of type β pro-collagen; PLT, platelet count; PT, prothrombin time; TB, total bilirubin; TC, total cholesterol; TG, triglycerides; TIMP-1, tissue inhibitor of metalloproteinase 1; β-NAG, N-acetyl β-glucosaminidase.
AAR, Williams, 1988	177	Mixed	AAR	n/a
PGA index, Poynard, 1991	624	Alcohol	PT, GGT, apoA1	n/a
PGAA index, Naveau, 1994	525	Alcohol	PT, GGT, apoA1, A2M	n/a
CDS index, Bonacini, 1997	75	HCV	PLT, AAR, PT	n/a
AP index, Poynard 1997	620	HCV	Age, PLT	0.763–0.690
BAAT score, Ratziu 2000	93	NAFLD	Age, BMI, ALT, TG	0.84
Fortunato, 2001	103	HCV	Fibronectin, prothrombin, ALT, PCHE, Mn-SOD, β-NAG	n/a
Pohl, 2001	211	HCV	AAR, PLT	n/a
FibroTest, Imbert-Bismut, 200111	339	HCV	A2M, Hpt, GGT, ApoA1, bilirubin	0.836–0.870
Kaul, 200212	264	HCV	PLT, AST, sex, spider nevi	n/a
Forns index, Forns, 200213	476	HCV	Age, GGT, cholesterol, PLT	0.86–0.81
APRI, Wai, 200314	270	HCV	AST, PLT	0.80–0.88
ELF-score, Rosenberg, 200415	1021	Mixed	Age, HA, PIIINP, TIMP-1	0.804
FIBROSpect II, Patel, 200416	696	HCV	HA, TIMP-1, A2M	0.831–0.823
FPI, Sud, 200417	302	HCV	Age, AST, TC, HOMA-IR, past alcohol intake	0.84–0.77
MP3, Leroy, 200418	194	HCV	PIIINP, MMP-1	0.82
HALT-C, Lok, 200519	1141	HCV	PLT, AAR, INR	0.78–0.81‡
Hepascore, Adams, 200520	221	HCV	Bilirubin, GGT, HA, A2M, age, sex	0.85–0.82
Fibrometer, Cales, 200521	383	Mixed	PLT, PI, AST, A2M, HA, urea, age	0.883–0.892
SHASTA index, Kelleher, 200522	95	HCV/HIV	HA,AST and albumin	0.878
Sakugawa, 200523	112	NAFLD	IV-7S, HA	n/a
Hui, 200524	235	HBV	BMI, PLT, albumin, TB, ALP	0.803–0.765
SLFG, Zeng, 200525	372	HBV	A2M, age, GGT, HA	0.84–0.77
FIB-4, Sterling, 200626	832	HCV/HIV	Age, AST, ALT, PLT	0.765§
Virahep-C, Fontana, 200627	399	HCV	age, AST, ALP, PLT	0.837–0.851
Mohamadnejad, 200628	276	HBV	HBV DNA levels, ALP, albumin, PLT,	0.91–0.85
FibroIndex, Koda, 200729	402	HCV	PLT, AST, γ-globulin	0.828–0.835
Alsatie, 200730	286	HCV	diabetes mellitus, PLT, AST, INR, bilirubin	0.79–0.75¶
Esmat, 200731	220	HCV	HA, age	0.84§
NAFLD fibrosis score, Angulo, 200732	733	NAFLD	Age, BMI, PLT, albumin, AAR, hyperglycemia	0.88–0.82

These panels are mainly based on two kinds of markers, direct and indirect. Direct markers are those directly linked to the modifications in ECM metabolism, such as hyaluronic acid and PIIINP. Indirect markers include a broad range of blood tests which have no direct link with liver fibrosis. They reflect liver dysfunction or other phenomena caused by fibrosis rather than fibrosis per se. Generally speaking, indexes including direct markers, such as the Fibrometer, may perform with greater accuracy, but indexes composed by only indirect markers are effective as well, and are usually more useful because they are based on routine blood tests that are easy to be performed in a general laboratory.

The diagnostic value of the models was assessed by calculating the area under the receiver operating characteristic curves (AUROC). Most studies reported an AUROC > 0.80 in differentiating significant fibrosis (fibrosis spread out the portal tract with septa) from no/mild fibrosis (no fibrosis or portal fibrosis without septa). Improved performance with a higher AUROC value was shown in differentiating between no cirrhosis and cirrhosis. But it must be underlined that the AUROC values in Table 2 each came from differently designed studies and are not suitable for making a comparison. Some well designed validation studies were done in the last 2 years, which may give us more reliable results.33,34

There are still some limitations of these marker panels to be considered. First, the design of every study differed in population characteristics, patient selection, significant fibrosis prevalence, blood test inclusion, biochemical measurement and liver histological assessment, which resulted in various panels with different markers and parameters. The agreement among these indexes is poor and a validation study is needed to choose a proper panel and cut-off value for clinical use. Second, none of the studies controlled for the degree of necro-inflammatory activity, most of the panels include markers likely to reflect or be affected by inflammation in the liver, which is much more mobile than fibrosis stage. Third, the formulas can easily fail because many markers included will be influenced by extrahepatic diseases or conditions such as inflammation, hemolysis, cholestasis, hypercholesterolaemia and renal failure. Finally, few of the studies include treated patients. It is not clear whether these indexes are suitable for assessing treatment response. However, a few studies by Poynard et al. suggested that FibroTest (BioLiveScale, Angers, France) could also be used as surrogate markers of the histological impact of treatments in patients infected by HCV and HBV.35

These indexes, in their current form, are not able to give us the exact stage of fibrosis in most studies. Their main value is to reduce the need for a liver biopsy by distinguishing significant fibrosis from no/mild fibrosis, and showing the presence of cirrhosis. It does not seem appropriate to completely replace liver biopsy with serum marker panels at the present time, but it can be anticipated that these indexes will become very useful in the clinical management of CLD by offering an attractive alternative to liver biopsy, as they are non-invasive, convenient, and inexpensive, and may allow the dynamic assessment of fibrosis. Validation in larger cohorts of patients with different CLD is needed before an index is proposed for extensive clinical use.

StephenW

Proteomics and glycomicsOver the last 2 years it was reported that the use of proteomic patterns in serum to distinguish individual stages of fibrosis could achieve perfect diagnostic sensitivity and specificity. Using a proteome-based fingerprinting model generated by surface-enhanced laser desorption/ionization time-of-flight (SELDI-TOF) ProteinChip (Ciphergen Biosystems, California, USA) arrays, Poon et al.36 achieved an AUROC of 0.93 in identifying significant fibrosis. Another proteomic index combining eight peaks established by Morra et al.37 could diagnosis advanced fibrosis with an AUROC of 0.88, significantly greater than the FibroTest AUROC of 0.81. Besides, The SELDI-TOF ProteinChip technology is useful for the early detection and prediction of HCC in patients with chronic HCV infection. Similar technologies have also been used to generate profiles of serum N-glycan profile for identifying liver fibrosis.38 Further studies identifying the altered peaks in these models to understand their origins may help to find new biomarks for fibrosis, or even improve our understanding in the mechanism of liver fibrosis.


RADIOLOGICAL TESTS
Since significant structural changes are present only in advanced CLD, routine examinations by ultrasound (US), computed tomography and magnetic resonance imaging (MRI) could produce specific findings, but with very limited sensitivity. Thus, persistent efforts have been made to search for technological developments.

Perfusion examinationsMR and Doppler US techniques are being studied to find sensitive perfusion changes in the progression of fibrosis. For example, the circulatory changes will result in a decrease of hepatic vein transit time (HVTT), which can be measured by microbubble-enhanced US. Using HVTT measurements, Lim et al. achieved 100% sensitivity and 80% specificity for diagnosis of cirrhosis, and 95% sensitivity and 86% specificity for the differentiation of mild hepatitis from more severe liver disease.39 Recent studies of hepatic microcirculation discovered that there is a progressive arterialization of the hepatic parenchyma during the progression of fibrosis, especially in the peripheral areas of the liver. The ‘central–peripheral phenomenon’ belonging to the transient hepatic arterial differences (THAD)40 can be detected in patients with advanced fibrosis or cirrhosis by computed tomography, which is a reliable qualitative measurement of liver perfusion changes. However, these techniques are not strong enough to distinguish precisely between different stages of fibrosis.


Liver stiffness measurementRecently, an important technological advance in fibrosis assessment was made by the FibroScan (Echosens, Paris, France), a new medical device based on one-dimensional transient elastography, which assesses fibrosis through liver stiffness measurement (LSM). A special probe generates an elastic shear wave propagating through the liver tissue: the harder the tissue, the faster the shear wave propagates. Transient elastography could accurately predict different stages of fibrosis or cirrhosis (AUROC: 0.79 for F ≥ 2, 0.91 for F ≥ 3, and 0.97 for F = 4. by the METAVIR scoring system).41

The major advantage of transient elastography compared with serum markers and marker panels is that it measures the liver directly and there is no interference from extrahepatic diseases or conditions. Furthermore, the test is standardized and completely non-invasive. Though assessing earlier fibrosis is the common shortcoming of various non-invasive tests, Colletta et al.42 reported that the agreement between transient elastography and a liver biopsy was much better than a FibroTest in normal transaminase HCV carriers with early stages of fibrosis.

Compared to a liver biopsy, transient elastography is painless, and rapid: it has no risk of complications and is, therefore, very well accepted. Transient elastography measures a volume of liver stiffness which is 100 times bigger than the biopsy specimen. The high reproducibility (the intra-observer and inter-observer agreement intra-class correlation coefficient was 0.9843) and acceptance of transient elastography makes it an attractive alternative to biopsy for individual follow-up.

There are also some physical limitations of transient elastography. The signal penetrates only 25–65 mm, making obesity (particularly the fatness of the chest wall) the most important cause of failure. But new technological developments may overcome this limitation. Additional limitations include a narrow intercostal space and ascites. However, the presence of ascites generally indicates cirrhosis by itself. Steatosis could have been expected to soften the liver because of the fat. Though current studies did not report that steatosis and necro-inflammation influence LSM,41 Fraquelli et al. found that transient elastography reproducibility is significantly reduced in patients with steatosis, an increased body mass index and lower degrees of hepatic fibrosis.43 So, larger cohorts with more severe grades of steatosis need to be studied before we make a conclusion.

The main reason that transient elastography cannot totally replace a liver biopsy is that it is only a means to stage disease. It is unable to diagnose liver disease by distinguishing between subtle diagnostic differences. Nor can transient elastography identify cofactors and comorbidities or grade necro-inflammation and steatosis. But it represents a totally different approach to assessing fibrosis and therefore could be combined with other non-invasive modalities to better assess liver fibrosis. The combined use of transient elastography and the FibroTest to evaluate liver fibrosis could avoid a biopsy procedure in most patients with chronic hepatitis C.44 Recently, Foucher et al. suggested that transient elastography might also be useful for predicting clinical complications of end-stage liver diseases such as esophageal varices and hepatocellular carcinomas,45 indicating the potential usefulness of transient elastography in the management of cirrhosis patients.


Real-time elastographyReal-time elastography is another ultrasound technique developed by Hitachi Medical Systems that can reveal the physical property of tissue using conventional ultrasound probes during a routine sonography examination. In the first study assessing real-time elastography for the detection of liver fibrosis,46 the AUROC was 0.75 for the diagnosis of significant fibrosis. Much higher diagnostic accuracy (AUROC = 0.93) was obtained by a mathematic combination of the elasticity score and two routine laboratory values (platelet count and γ-glutamyltransferase), which provided a superior way to combine serological and radiological tests together.


Magnetic resonance elastographyMagnetic resonance elastography (MRE) is a technique using a modified phase-contrast MRI sequence to image propagating shear waves in tissue. The technique has been previously applied to quantitatively assess the viscoelastic properties of the breast, brain and muscle in humans. Several recent studies showed that MRE is also a feasible method for assessing the stage of liver fibrosis.47 MRE has several potential advantages compared with ultrasound transient elastography. It can be performed in obese patients. It can assess larger volumes and provide full three-dimensional information about the viscoelastic parameters of tissues. With MR techniques a comprehensive examination of the liver can be performed, including MRE, contrast-enhanced MRI to detect hepatocellular carcinomas and perfusion MRI to assess liver function. However, direct comparison of the two methods is still not available. More studies are needed to define the sensitivity and specificity of this new technique.


Double contrast material-enhanced MRIThe conspicuity of gadolinium-enhanced lesions is increased in the setting of a decreased signal intensity from an uninvolved liver parenchyma following a superparamagnetic iron oxide injection. This MRI technique has been used to improve detection of focal hepatic lesions and hepatocellular carcinomas. Recently, Aguirre et al.48 examined 101 CLD patients who underwent double-enhanced MR imaging to detect hyperintense reticulations, which are postulated to represent septal fibrosis. They achieved an accuracy of greater than 90% for the diagnosis of fibrosis compared with histopathological analysis.


Diffusion weighted magnetic resonance imagingDiffusion weighted magnetic resonance imaging (DWMRI) has been widely used in brain imaging for the evaluation of acute ischemic stroke. With the advent of the echo-planar MRI technique, it became possible to be applied in the abdomen for the characterization of focal hepatic lesions. Recently, using DWMRI to measure the apparent diffusion coefficient (ADC) of water, a parameter that is dependent on the tissue structure, has been introduced in the assessment of liver fibrosis. The ADC value is lower in livers with heavier fibrosis because of the restriction of water diffusion in fibrotic tissue. Lewin et al. assessed the performance of DWMRI in 54 patients with chronic HCV infection with reference to several other non-invasive methods.49 In discriminating significant fibrosis, the area under the curve (AUC) values were 0.79 for DWMRI, 0.87 for transient elastography, 0.68 for the FibroTest, 0.81 for aspartate aminotransferase to platelet ration index, 0.72 for the Forns index and 0.77 for hyaluronate. DWMRI performed better in discriminating between patients staged F3–F4, when the AUC value increased to 0.92, the same as in transient elastography. But besides fibrosis, it seems that ADC values might also reflect the intensity of inflammation, necrosis and steatosis. However, DWMRI still benefits from the intrinsic advantages of MRI. Several other MR techniques have also been introduced in the area of fibrosis assessment, such as ultra-short echo time MRI50 and magnetic resonance spectroscopy,51 but the research data are still insufficient.



CONCLUSION
The increase of potentially effective management for CLD such as antiviral and antifibrotic therapies has led to an urgent need for a rapid, safe and repeatable tool to assess fibrosis of CLD and to follow up progression or regression of fibrosis during treatment. A liver biopsy has been the gold standard for the assessment of hepatic fibrosis, but the invasive procedure has considerable limitations and fails to satisfy current needs. Many non-invasive methods have been proposed with the aim of substituting for a liver biopsy. The numerous advances in serological and radiological techniques and their combinations have allowed clinicians to satisfactorily identify patients without resorting to a liver biopsy. But each has some deficiencies and a liver biopsy will still have an important role to play. Applying new techniques for the detection of fibrosis may potentially circumvent the pitfalls and deficiencies of existing surrogates mentioned above. However, further studies are needed to develop or validate non-invasive tests that can accurately reflect the full spectrum of hepatic fibrosis in CLD. But an incorrigible defect in our studies will be the questionable gold standard we have to use. Biopsy failure is more common than diagnostic failure of markers.52 Mathematical modeling suggests that, assuming either 80% or 90% diagnostic accuracy of a liver biopsy, non-invasive tests cannot achieve an AUROC better than 0.9 and are likely to perform between 0.75 and 0.9, exactly where they are today. We may find a better surrogate for a liver biopsy, but how we can prove it will be a challenge. A laparoscopic biopsy can decrease sampling error and increase the reliability of a histopathological assessment. Using automated image analysis to assess texture features and the shape representation of the fibrosis structural expansion can turn the current semiquantitative methods of liver fibrosis assessment into real quantitative methods with significant reduction in variability and subjectivity.53 Validating non-invasive tests against not only histological stage scores but also digital image analysis and clinical outcomes may also be a better choice.


FINANCIAL SUPPORT
This study was supported by the Prominent National Projects of Science & Technology (No: 2008ZX10203), the National High Technology Research and Development Program of China (863 Program, No: 2006AA02A411), Science and Technology Commission of Shanghai Municipality (No: 064119519), and Shanghai Leading Academic Discipline Project (No: Y0205).

lin12345