Noninvasive assessment of liver fibrosis

Authors

  • Stella M. Martínez,

    1. Liver Unit, Hospital Clínic, IDIBAPS (Institut d'Investigacions Biomèdiques August Pi i Sunyer) and CIBERehd (Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas), Barcelona, Spain
    Search for more papers by this author
  • Gonzalo Crespo,

    1. Liver Unit, Hospital Clínic, IDIBAPS (Institut d'Investigacions Biomèdiques August Pi i Sunyer) and CIBERehd (Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas), Barcelona, Spain
    Search for more papers by this author
  • Miquel Navasa,

    1. Liver Unit, Hospital Clínic, IDIBAPS (Institut d'Investigacions Biomèdiques August Pi i Sunyer) and CIBERehd (Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas), Barcelona, Spain
    Search for more papers by this author
  • Xavier Forns

    Corresponding author
    1. Liver Unit, Hospital Clínic, IDIBAPS (Institut d'Investigacions Biomèdiques August Pi i Sunyer) and CIBERehd (Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas), Barcelona, Spain
    • Liver Unit,Villarroel 170, Hospital Clinic, Barcelona 08036, Spain
    Search for more papers by this author
    • fax: (34)-93-451-55-22


  • Potential conflict of interest: Nothing to report.

Abstract

Liver biopsy has long been an important tool for assessing the degree of liver fibrosis. Information on the presence and degree of liver fibrosis is useful before making therapeutic decisions or predicting disease outcomes. The need to stage liver fibrosis, however, should decrease as treatment options become more successful (as has occurred with viral hepatitis). In recent years, noninvasive tests have demonstrated a reasonable ability to identify significant fibrosis, cirrhosis in particular, nor is it surprising that liver disease specialists and patients favor a noninvasive approach. However, only those tests with the highest diagnostic accuracy, cost-effectiveness, and availability should be implemented. Apart from their diagnostic accuracy, the potential ability of these tests to predict disease outcomes (a more relevant endpoint) should be compared with that of liver biopsy. Indeed, the use of a standardized system to evaluate the utility of biomarkers would facilitate their implementation in clinical practice. (HEPATOLOGY 2011.)

Noninvasive Assessment of Liver Fibrosis in the 21st Century

The tremendous strides made in biomedicine and biotechnology during the last decade have led to novel, rapid, and noninvasive methods that challenge long-standing “gold standard” invasive diagnostic and therapeutic procedures. Just a few examples include noninvasive coronary angiography using computed tomography (CT), cardiac magnetic resonance imaging (MRI) for the noninvasive evaluation of cardiomyopathies, and high-resolution CT scanning for diagnosis of interstitial lung diseases.

Noninvasive diagnosis of liver fibrosis is one of the fields that has evolved most rapidly in recent years. Diagnosis and follow-up of liver diseases has long relied on liver biopsy, and only recently has its value as a method to assess the severity of liver diseases (or to follow-up disease progression) been questioned. Although biopsy is used to stage most cases of liver disease, it is well known that this procedure has several limitations. First, liver biopsies only sample an extremely small portion of the liver (1/50,000) and therefore, sampling errors can occur, especially when smaller sized biopsies are analyzed.1, 2 In addition, histological examination is prone to intraobserver and interobserver variation, which may occur even when widely validated systems are used to score liver damage. Finally, liver biopsy is an invasive procedure with associated morbidity: pain occurs in 20% of patients and major complications (such as bleeding or hemobilia) in 0.5%.2 For this reason, liver biopsy has poor tolerance, particularly if it needs to be repeated over time in an individual patient. Moreover, patients are now aware of the availability of noninvasive methods to assess the degree of liver damage and are thus reluctant to undergo an invasive procedure.

Despite the aforementioned limitations, there has been a clear resistance to accept noninvasive diagnosis of liver fibrosis as a viable and preferable alternative to liver biopsy. The reasons for this are various. First, there is a paucity of well-designed studies assessing noninvasive methods, and sufficient external validation for some of the proposed methods is lacking. Second, the number of proposed methods to assess disease severity remains in a state of constant growth and there is practically no time to validate or test them all. Third, liver biopsy itself is not an ideal gold standard. Finally, and not least important, there is still significant opposition to changing what has long stood as dogma. The aforementioned reasons may explain why the introduction of noninvasive methods in clinical practice is making such slow headway in the field of hepatology. An exception to this rule is France, where three well-validated methods (Fibrotest, Fibrometer, and FibroScan) have been approved by the public health system and are routinely used in clinical practice.

Abbreviations

ALT, alanine aminotransferase; ARFI, acoustic radiation force impulse; APRI, AST-to-platelet ratio index; AST, aspartate aminotransferase; AUROC, area under the receiver operating characteristic curve; BMI, body mass index; CT, computed tomography; CTP, Child-Turcotte-Pugh; ELF, Enhanced Liver Fibrosis panel; GGT, gamma glutamyl transpeptidase; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; MELD, Model for End-Stage Liver Disease; MRI, magnetic resonance imaging; NAFLD, nonalcoholic fatty liver disease; PBC, primary biliary cirrhosis; TE, transient elastography; TIMP, tissue inhibitor of metalloproteinase.

Fibrosis Staging in Liver Diseases: How and What for?

In slowly progressive diseases such as chronic hepatitis, liver fibrosis staging based on a liver biopsy has long been the gold standard for assessing disease severity and for establishing outcomes. Any new method needs to be compared with liver biopsy, which has a number of limitations. When assessing the diagnostic accuracy of various noninvasive tests, the relatively high misclassification rate of the gold standard (liver biopsy) makes it impossible to achieve a perfect concordance (area under the receiver operating characteristic curve [AUROC] ∼1). Assuming a sensitivity and specificity of liver biopsy >90% (the best-case scenario) and a prevalence of significant fibrosis ∼40%, a “perfect” noninvasive test would only reach an AUROC of 0.90.3 Another consideration is the fact that fibrosis staging is defined by architectural changes, which is different than the total amount of fibrosis. Thus, presumption of a direct correlation between fibrosis stages and the values obtained with a noninvasive test may be erroneous.4 Despite the good correlation between fibrosis staging and specific outcomes (time to develop cirrhosis, time to develop clinical decompensation), the progression of liver fibrosis over time is commonly not linear and can be influenced by many variables.5, 6 The speed of fibrosis progression is much more important than establishing a static diagnosis. Finally, standard histological methods do not allow for any determination as to whether fibrosis is under the process of deposition-degradation or if it is an established process (e.g., as a consequence of a previous liver injury).

Despite the aforementioned limitations, most studies rely on liver biopsies as the gold standard to evaluate the performance of a noninvasive method (which, as stated above, represents an inherent limitation). A few studies published recently have used the measurement of portal pressure as the gold standard to assess disease severity7,8: the method is reproducible and is not subject to sampling error. In certain situations (e.g., patients with advanced liver disease, patients who have undergone liver transplantation) its measurement should be taken for assessing disease severity, especially because it offers an excellent ability to predict clinical endpoints. However, portal pressure is not a marker of the liver fibrosis stage and can neither detect nor discriminate between the initial stages of fibrosis.

Most well-validated noninvasive methods have shown good diagnostic accuracy in identifying patients with a significant degree of fibrosis (i.e., fibrosis expanding beyond the portal tract). Sensitivities and specificities above 85% can be considered sufficient due to the lack of relevant clinical consequences for false positive and false negative cases. In addition, noninvasive tests can be repeated over time, and in cases of indeterminate results, two or more methods can be combined. Although the identification of significant fibrosis (≥F2) has been regarded as an important target (particularly in the field of viral hepatitis), its real value as a static measure of disease severity is arguable. The future incorporation of more effective therapies (especially for hepatitis C) will most likely change our current approach to chronic hepatitis.

The identification of individuals with bridging fibrosis or liver cirrhosis is more critical: in such cases, sensitivity must be very high, not only due to the potential indication of a specific treatment, but also because screening for hepatocellular carcinoma (HCC) and gastroesophageal varices is mandatory in patients with liver cirrhosis. Fortunately, the performance of noninvasive methods to assess the presence of cirrhosis is excellent (see below).

Noninvasive Methods for Assessing Liver Fibrosis

The first approach used to assess the degree of liver fibrosis by noninvasive means consisted of routine biochemical and/or hematological tests. These so-called class II biomarkers are indirect serum markers and are based on the evaluation of common functional alterations in the liver, alterations that do not necessarily reflect extracellular matrix turnover and/or fibrogenic cell changes. More recently, a better understanding of the pathophysiology of liver fibrosis has prompted many investigators to use more refined markers to identify different fibrosis stages. The latter, termed class I biomarkers, are those intended to detect extracellular matrix turnover and/or fibrogenic cell changes. Finally, the extraordinary advances in imaging techniques have allowed the application of new diagnostic tools to assess fibrosis in the liver (Fig. 1).

Figure 1.

Noninvasive assessment of liver fibrosis. Most of the noninvasive methods used currently have a good accuracy to identify advanced stages of fibrosis (which is of clinical relevance). In the near future, it is possible that the incorporation of genetic, proteomic, and metabolomic profiles will allow us to identify liver fibrosis at its early stages. SNP, single-nucleotide polymorphism.

Only a few years ago, data on the diagnostic performance of noninvasive fibrosis markers was restricted to patients with chronic hepatitis C, where the information was useful to adopt therapeutic decisions. In the last few years, interest in this field has expanded to other liver diseases and, more importantly, noninvasive assessment of liver fibrosis has recently shown its potential for monitoring the natural history of liver disease and predicting specific clinical outcomes.

Routine Serological Markers.

Generally, class II biomarkers are identified from retrospective studies in which combinations of routine laboratory tests associated with liver function are tested for their predictive value to identify significant fibrosis or cirrhosis. Some of these scores have been widely validated in large cohorts of patients9-15 (Table 1; Supporting Table 1). The combination of routine laboratory tests with fibrosis biomarkers may increase the accuracy for identifying cases of significant fibrosis and cirrhosis.9, 10, 16 Although identification of individual fibrosis stages is still not possible, some of the published scores do allow the classification of 50%-70% of individuals (as having either significant or nonsignificant fibrosis) with high positive and negative predictive values. Their major drawback is that they are not sensitive enough to identify patients with mild degrees of fibrosis but who are at risk of progression.

Table 1. Performance of Noninvasive Methods to Assess Liver Fibrosis in Patients With Viral Chronic Hepatitis C*
Score (Original Reference)Serum Markers/FibroscanEtiologyn≥ F2 (%)AUC ≥ F2F4 (%)AUCF4
  • *

    The number of patients (n), the prevalence significant fibrosis (≥F2) and cirrhosis (F4) and the AUC values are based on published studies including >100 patients.

  • Two detailed tables with the results of individual studies are included as supporting material (Supporting Tables 1 and 2).

  • Not all studies assessed AUC for F4.

  • AUC, area under the curve; LT, liver transplantation.

Fibrotest (12)GGT, haptoglobin, bilirubin, apolipoprotein A1, alpha-2-macroglobulinHCV234233-740.74-0-893-250.82-0.92
  HBV21856-680.77-0.8515-200.76-0.87
Forns (11)Age, GGT, cholesterol, plateletsHCV198232-590.75-0.913-20-
  HBV45656-740.63-0.7215-260.81
APRI (14)AST, plateletsHCV316027-740.69-0.883-250.61-0.94
  HBV88656-790.7215-200.64-0.76
FIB-4 (13)Age, ALT, AST, plateletsHCV (HIV)177821-360.74-0.8570.91
  HBV77656-790.74-0.8615-340.80-0.93
Hepascore (9)Age, sex, alpha-2-macroglobulin, hyaluronate, bilirubin, GGTHCV166039-790.74-0.866-340.80-0.94
Fibrometer (10)Platelets, prothrombin time, macroglobulin, AST, hyaluronate, age, ureaHCV103941-560.78-0.894-150.94
  HBV108560.74150.89
ELF (15)N-terminal propeptide of collagen type III, hyaluronic acid, TIMP-1, ageHCV/HBV134627-640.77-0.8712-160.87-0.90
Fibroscan (19)Transient elastographyHCV205237-740.72-0.918-250.87-0.98
  HCV-LT47036-680.81-0.909-170.87-0.98
  HBV81650-680.80-0.938-25093-0.96

Fibrosis Biomarkers.

The identification of hepatic stellate cells has been a key development in understanding the wound-healing response in the liver. Activation of hepatic stellate cells is the dominant event in hepatic fibrogenesis, which is characterized by the transformation of quiescent cells into proliferative, fibrogenic, and contractile myofibroblasts. Fibrosis progresses when there is an imbalance between extracellular matrix degradation and production.17 Although liver fibrosis is a local reaction of the liver to chronic injury, serum levels of fibrogenic cytokines, extracellular matrix proteins, and degradation products are markedly increased in cases of advanced fibrosis (bridging fibrosis or cirrhosis). Thus, there is a growing interest in the identification of patients with advanced fibrosis by serum markers. The most common markers used in current assays involve measuring products of extracellular matrix synthesis or degradation and the enzymes that regulate their production or modification: hyaluronic acid, serum collagenases and their inhibitors (tissue inhibitor of metalloproteinase [TIMP]), and profibrogenic cytokines (such as transforming growth factor β1)15, 17 (Table 1). The quantification of peptide epitopes generated from enzymatic cleavage of proteins during extracellular remodeling may aid in the discovery of new liver fibrosis markers.

Some limitations inherent to fibrosis biomarkers are their lack of sensitivity during the initial stages of liver fibrosis and their lack of specificity: they can detect fibrogenesis in organs other than the liver and can be affected by renal or liver failure. The fact that some of the markers are not routinely available in most clinical laboratories will prove less of a problem in the near future, because some of the assays based on fibrogenesis markers have been patented and will soon be commercialized.

Imaging Techniques.

Ultrasonography, CT scan, or MRI have traditionally been used to explore the liver. These methods are able to detect changes in the liver parenchyma when there is significant fibrosis (bridging fibrosis and mainly cirrhosis) and signs of portal hypertension (enlarged spleen, collateral venous circulation, enlarged portal vein). However, these methods are not useful for identifying patients with less advanced stages of fibrosis. Recently, optical analysis of CT images of the liver (Fibro-CT) has been used to assess fibrosis in patients with chronic hepatitis C virus (HCV) infection.18 The use of fibro-CT, however, is time-consuming and more expensive than the noninvasive serum markers currently in use.

Transient elastography (TE) is probably the most widely used noninvasive method in Europe for assessing the degree of liver fibrosis (FibroScan).19 The concept is simple: a vibration of mild amplitude and low frequency is transmitted to the tissue, which induces an elastic shear wave that propagates within the liver. Pulse-echo ultrasonic acquisitions follow the shear wave and measure its speed (Fig. 2A). The velocity of wave propagation relates directly to tissue stiffness (the harder the tissue, the faster the shear propagates), which is measured in kilopascals. The advantages of this method are: (1) it is rapid, noninvasive, and reproducible; (2) it acquires information from a much larger portion of the tissue compared with liver biopsy, and therefore, the risk of sampling error is significantly lower; (3) it can be used in different liver diseases. TE has some limitations, however. It has proven unsuccessful in individuals with narrow intercostal spaces and morbid obesity, and increased liver stiffness is not always a surrogate of fibrosis (the presence of significant necroinflammation or extrahepatic cholestasis may increase liver stiffness values in the absence of fibrosis). Despite the aforementioned limitations, several studies have already evaluated the accuracy of TE in identifying patients with significant fibrosis or cirrhosis7, 20-24 (Table 1; Supporting Table 2). The diagnostic accuracy is sufficiently good in identifying significant fibrosis (particularly if the underlying disease is taken into account) and is excellent in identifying liver cirrhosis.24, 25

Figure 2.

(A) TE and (B) ARFI obtained in two patients with chronic hepatitis C. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Table 2. Scores Combining Routine Laboratory Tests and Fibrosis Biomarkers to Assess Liver Fibrosis in Alcoholic Liver Disease
Score (Reference)Variables
PGAA (38)Prothrombin time, GGT, apolipoprotein A1, and alpha-2 macroglobulin
APRI (14)AST, platelets
Fibrotest (12)GGT, haptoglobin, bilirubin, apolipoprotein A, alpha-2-macroglobulin
Hepascore (9)Age, sex, alpha-2-macroglobulin, hyaluronate, bilirubin, GGT
Fibrometer (10)Platelets, prothrombin time, macroglobulin, AST, hyaluronate, age, urea

Recently, acoustic radiation force impulse (ARFI) imaging technology has been implemented as a valid method to assess liver fibrosis. ARFI imaging permits evaluation of the elastic properties of a region of interest while performing a real-time B-mode conventional hepatic ultrasonography. Results are expressed in meters per second and the region of interest can be chosen by means of ultrasound guidance, thus avoiding large blood vessels or ribs (Fig. 2B). Recent studies have shown excellent diagnostic accuracy in identifying significant fibrosis and cirrhosis in patients with various liver diseases.26, 27 Its inclusion in a conventional ultrasound machine may offer an advantage in some centers. However, studies with higher numbers of patients under different settings are needed in order to successfully incorporate this promising new tool into clinical practice.

Several technological advances have been made in the clinical application of liver MRI: contrast-enhanced MRI, diffusion-weighted MRI, and magnetic resonance elastography.28 The latter uses a modified phase-contrast method to image the propagation characteristics of the shear waves within the liver. The potential assessment of the entire liver parenchyma, the lack of an acoustical window requirement, and operator independence are clear advantages. In addition, this method may aid in quantifying hepatic fat content. The drawbacks of magnetic resonance elastography are its cost and the fact that it is time-consuming. Recently, the incorporation of proton magnetic resonance spectroscopy (which allows for the analysis of molecular tissue composition) appears to be a safe and reproducible tool for assessing hepatic fat content. In contrast to liver tissue examination (in which the percent of hepatocytes with fat droplets are assessed), this method determines the volume fractions of lipids; livers affected by fat infiltration exhibit an increase in the intensity of the lipid resonance peak.

Noninvasive Assessment of Liver Fibrosis in Different Clinical Settings

Chronic Hepatitis.

Most of the published studies addressing the use of serum markers for the assessment of liver fibrosis have been performed in patients with chronic hepatitis, particularly in those with chronic hepatitis C. Several of these methods have been extensively validated in large cohorts of patients and are widely used in routine clinical practice9-15 (Table 1; Supporting Table 1). A similar situation occurs with TE.7, 20-24, 29 This method has been validated in large cohorts of individuals with chronic HCV infection and it has shown high diagnostic accuracy in identifying individuals with significant fibrosis and, in particular, liver cirrhosis (Table 1; Supporting Table 2).

The use of two or more noninvasive methods increases the diagnostic accuracy of an individual assay.30-33 Castera et al.22 showed that the combination of TE and Fibrotest was very useful for diagnosing significant fibrosis (≥F2) and cirrhosis (F4), with areas under the ROC curve measuring 0.88 and 0.95, respectively. Sebastiani and coworkers34 investigated the viability of combining the aspartate aminotransferase (AST)-to-platelet ratio index (APRI) with Fibrotest-Fibrosure (Sequential Algorithm for Fibrosis Evaluation [SAFE]) to identify significant fibrosis and cirrhosis in more than 2000 patients with chronic hepatitis C. The algorithm used to identify significant fibrosis avoided 50% of the liver biopsies, whereas an algorithm used to identify cirrhosis avoided more than 80% of the biopsies. The rates of discordance between the histological findings and SAFE were low (approximately 10%) and in most cases, TE confirmed the results of the noninvasive test.

In the case of patients with chronic hepatitis B virus (HBV) infection, the number of studies assessing noninvasive markers is smaller. Sebastiani et al.35 assessed the utility of several noninvasive markers in a cohort of 110 patients with chronic hepatitis B who had undergone liver biopsy. Fibrotest and APRI achieved the highest diagnostic accuracy in identifying patients with significant fibrosis, whereas a combination of both tests in a single algorithm had an accuracy of 97%; a second algorithm incorporating Fibrotest and APRI and geared toward identifying liver cirrhosis obtained similar results. In yet another study, the combination of the Forns score with TE increased the diagnostic accuracy in distinguishing individuals with chronic HBV infection and significant liver fibrosis.36 Other noninvasive scores have recently been tested in patients with chronic HBV infection and have obtained similarly good results (Supporting Table 1). In clinical practice, it is important to consider the potential influence of significant increases in AST and alanine aminotransferase (ALT) values when interpreting the results of TE.37

Alcoholic Liver Disease.

Alcoholic liver disease is a major public health problem and is characterized by a wide spectrum of liver damage, ranging from hepatic steatosis to cirrhosis and including alcoholic hepatitis. The distribution of liver fibrosis is different from that of patients with chronic hepatitis C (perisinusoidal and perivenular fibrosis is frequent in individuals with alcoholic liver disease). In addition, necroinflammation and steatosis may be significant in individuals with active alcohol consumption. The latter histological features may influence the results of some commonly used noninvasive methods, such as TE. Assessment of the degree of liver fibrosis in individuals with alcoholic liver disease has prognostic implications and thus, noninvasive assessment of fibrosis remains important.

Several noninvasive laboratory tests have been used in the past to assess liver fibrosis in patients with alcoholic liver disease, including the PGA (prothrombin time, gamma glutamyltransferase, apolipoprotein A1) and the PGAA (PGA + α2-macroglobulin) indexes, hyaluronic acid, Fibrometer, and Hepascore9, 10, 12, 14, 38 (Table 2). Naveau et al.39 recently analyzed the accuracy of different serum biomarker–based tests (Fibrotest, FibrometerA, Hepascore) to assess liver fibrosis and to determine their prognostic value in patients with alcoholic liver disease. More than 200 such patients who had undergone liver biopsy were followed for a median time of 8 years. All three tests demonstrated excellent diagnostic accuracy in identifying advanced fibrosis and cirrhosis; more importantly, Fibrotest and fibrosis stage at study entry time were independently associated with survival.

In another recent study, Nguyen-Khac et al.40 assessed liver fibrosis by TE and several serological assays in 103 alcoholic patients. The tested serological assays included Fibrotest, Fibrometer, Hepascore, APRI, PGA, PGAA, and hyaluronic acid (Table 2). Using the METAVIR fibrosis staging system, TE showed the highest accuracy in distinguishing different stages of fibrosis. The higher cutoff values obtained when identifying significant fibrosis and cirrhosis obtained in this cohort (compared to patients with chronic hepatitis C) may be explained by the particular distribution of liver fibrosis in alcoholic patients, which emphasizes the need for disease-specific cutoff values. The results of this and other studies have proven similar41 and support the utility of TE in alcoholic patients. Validation in large and independent series is still needed before recommending its use in routine clinical practice.

Nonalcoholic Fatty Liver Disease and Nonalcoholic Steatohepatitis.

Nonalcoholic fatty liver disease (NAFLD) is a major cause of liver disease, stemming from the increasing prevalence of obesity and type 2 diabetes in the Western World. As with other liver diseases, the NAFLD spectrum is wide and may range from simple liver steatosis to steatohepatitis, cirrhosis, and HCC. Although most patients diagnosed with NAFLD are at the “mild end” of the disease spectrum, identifying those who have already developed liver fibrosis is relevant for prognostic considerations. In addition, such patients may be candidates for emerging treatments. As with other diseases in this field, interest in applying noninvasive methods to assess liver fibrosis in patients with NALFD has increased in recent years.42

In patients with nonalcoholic steatohepatitis (NASH), the variables more frequently associated with significant fibrosis are age, diabetes, increased insulin resistance (Homeostasis Model Assessment [HOMA]), AST/ALT ratio, platelet count, body mass index (BMI), and hyaluronic acid.43 Several scores combining various markers have been specifically tested in the context of NAFLD44-48 (Table 3). In general, all of these noninvasive markers perform well in detecting advanced liver fibrosis, although their negative predictive value for the presence of mild to moderate fibrosis is low.

Table 3. Scores Combining Routine Laboratory Tests and Fibrosis Biomarkers to Assess Liver Fibrosis in Patients With NAFLD/NASH
Score (Reference)Variables
BAAT (44)Age, BMI, ALT, triglycerides
Fibrotest (45)Alpha-2-macroglobulin, apolipoprotein A1, haptoglobin, bilirubin, GGT
NAFLD fibrosis score (46)Age, BMI, platelets, albumin, AST/ALT, IFG / diabetes
ELF (47)N-terminal propeptide of collagen type III, hyaluronic acid, TIMP-1

Recently, the utility of the Enhanced Liver Fibrosis panel (ELF) has been validated in an independent cohort of 196 patients with NAFLD.47 The ELF panel had an accuracy of 0.90 (measured as AUROC) in identifying individuals with severe fibrosis, 0.82 for those with moderate fibrosis, and 0.76 for those without fibrosis. Importantly, the addition of some simple variables (such as BMI, the presence or absence of diabetes, platelet count, AST/ALT ratio) considerably improved the accuracy of the ELF score. More than 80% of liver biopsies might be avoided by applying this score in patients with NAFLD. This test may also prove useful in pediatric NAFLD.49

TE has also shown good diagnostic accuracy in identifying the presence of fibrosis in individuals with NAFLD50, 51 (Supporting Table 1). In a study by Wong et al., which included more than 200 patients, liver stiffness was not affected by hepatic steatosis, necroinflammation, or BMI, and discordances between liver biopsy and TE were correlated to small tissue size and absent or mild fibrosis.

Cholestatic Liver Diseases and Metabolic Liver Diseases.

Although most published studies assessing TE have been made in patients with chronic hepatitis, there are few reports on its potential use in other chronic liver diseases.23, 52 Corpechot et al.23 evaluated TE in 101 patients with chronic cholestatic diseases (mostly primary biliary cirrhosis [PBC] and a few with primary sclerosing cholangitis): liver stiffness measurements very accurately identified patients with advanced fibrosis and cirrhosis (Supporting Table 2). This is relevant because the prognosis of chronic cholestatic diseases depends, at least in part, on the extent of fibrosis in the liver. Most serum fibrosis scores developed in patients with chronic hepatitis C are not appropriate for assessing PBC or primary sclerosing cholangitis because variables included in such scores (i.e., gamma glutamyl transpeptidase [GGT], cholesterol, or apolipoprotein A1) are influenced by cholestasis.

The diagnostic performance of the ELF assay has recently been validated in a large cohort of patients with PBC.53 Serial liver biopsies and serum samples were collected regularly from 161 individuals with PBC over a period of 7 years. The authors explored the accuracy of the score in distinguishing patients with significant fibrosis and cirrhosis, and more importantly, in forecasting clinical outcomes. Clinical progression was defined as the appearance of esophageal varices, clinical decompensation, liver transplantation, or death. The correlation of ELF with histology was acceptable, with an AUROC of 0.76 in predicting cirrhosis (Ishak stage = 5 or 6) and 0.75 in forecasting significant fibrosis (Ishak stage = 3-6). Nevertheless, the most relevant result of this study was the accuracy of baseline ELF in predicting clinical outcomes (see below).

In the context of metabolic liver diseases, a recent study evaluated the performance of serological markers in identifying advanced fibrosis in patients with hereditary hemochromatosis.54 Among the markers included in this study were platelet counts, ferritin levels, AST levels, and direct fibrogenesis markers (hyaluronic acid and collagen type IV). Despite the small sample size of the study (n = 56), high hyaluronic acid concentration (>46.5 ng/mL) proved 100% sensitive and specific at identifying patients with cirrhosis in individuals with ferritin levels >1000 μg/L.

Liver Transplantation and Hepatitis C Recurrence.

The use of liver fibrosis serum markers (particularly when based on indirect fibrosis markers) is problematic in patients undergoing liver transplantation. Some of the variables included in the scores (ALT, platelet counts, cholesterol) may vary due to causes unrelated to the deposition of collagen in the liver graft. A recent study, however, has shown how the ELF score can discriminate, early after the liver transplantation procedure, between patients with mild and progressive hepatitis C recurrence.55

TE appears to be a promising tool in this setting. There are several studies that have shown TE to be very accurate in identifying patients with advanced fibrosis and portal hypertension in the context of hepatitis C recurrence after liver transplantation.7, 56-58 In the study carried out by Carrión,7 124 liver transplant recipients with HCV infection underwent 169 liver biopsies and 129 liver hemodynamic analyses, paired with liver stiffness measurements. As expected, the presence of portal hypertension (hepatic venous pressure gradient ≥6 mm Hg) was indicative of advanced liver fibrosis (≥F2). Importantly, there was a good correlation between portal pressure and liver stiffness; the area under the ROC curve for diagnosing portal hypertension was 0.93. Given the need for frequent histological assessment in liver transplant patients with HCV infection recurrence, TE appears to be an extremely useful tool for follow-up of disease progression in such a population.58 This is particularly relevant because the presence of significant fibrosis or increased portal pressure can identify those patients at risk of clinical decompensation.59

Noninvasive Markers of Liver Fibrosis for Assessing Disease Outcome and Response to Therapy

The most common liver diseases (viral hepatitis, alcoholic liver disease) can be diagnosed without the need of a liver biopsy. A liver biopsy is used to exclude coexisting problems (e.g., steatosis, granulomas) or to assess any architectural changes in the liver (i.e., fibrosis) in order to predict disease outcome. Recent data suggest that noninvasive tests can also play a role in identifying those patients at risk of disease progression (e.g., clinical decompensation, HCC, and liver-related death). Importantly, the latter can be repeated over time for patient follow-up (Fig. 3).

Figure 3.

(A) Longitudinal assessment of APRI and Forns' scores in a patient with chronic hepatitis C who achieved sustained virological response after antiviral therapy (pegylated interferon [Peg-IFN] and ribavirin [RBV]) (•) and a patient who received the same therapy but did not respond to treatment (▪). (B) Assessment of TE in three HCV-infected liver transplant recipients: one with mild hepatitis C recurrence (▪), a second with severe hepatitis C recurrence who did not respond to antiviral treatment (▴), and a third with severe hepatitis C recurrence and sustained virological response (•). Dotted horizontal line represents a liver stiffness value of 8.7 kPa (cutoff point for F2 fibrosis). Follow-up is shown in years. OLT, orthotopic liver transplantation; TE, transient elastography.

The potential role of TE to assess clinical outcomes is supported by its good correlation with portal pressure, which accurately predicts clinical events. A study that included 165 patients with cirrhosis demonstrated that liver stiffness values below 19 kPa were highly predictive of the absence of large varices, with a negative predictive value of 93%.60 Other studies have reached similar conclusions.61 It is important to remember, however, that the correlation between portal pressure and liver stiffness decreases when hepatic venous pressure gradient values exceed 12 mm Hg.7, 8, 62 Liver stiffness measurement may also be useful in identifying patients at risk of developing HCC. In a recent Japanese study,63 866 HCV-positive patients underwent liver stiffness measurement at baseline and were followed for a mean period of 3 years. HCC developed in 77 patients. The 3-year cumulative probability of developing HCC correlated to a significant degree with baseline liver stiffness, with rates as low as 0.4% in patients with values ≤10 kPa and as high as 38% in those with values >25 kPa. Both studies may have clinical implications in terms of screening for varices and HCC, but the results need validation in a larger cohort of patients.

Noninvasive fibrosis markers may also be useful for predicting survival. The prognostic values of Fibrotest, FibrometerA, and Hepascore were recently compared in a cohort of 218 patients with alcoholic liver disease with a median follow-up of 8 years.39 All patients were heavy alcohol consumers and had undergone a liver biopsy. Interestingly, all three panel biomarker tests achieved the same performance (comparable to liver histology) for the study endpoint (survival and liver-related death): the AUROC ranged from 0.77-0.80. In a completely different setting, Mayo and collaborators53 showed that the ELF score was capable of identifying patients suffering PBC who were at risk of clinical decompensation. In this study, serial liver biopsy specimens and serum samples were collected over time (median 7 years) in a large cohort of patients with PBC. The probability of developing a clinical event (appearance of varices, clinical decompensation, liver transplantation, or liver-related death) correlated to a significant degree with the baseline ELF score, and proved quite small (8%) in individuals with a low score, compared to 46% in those with a high baseline score. Similar results were obtained in a HALT-C (Hepatitis C Antiviral Long-Term Treatment Against Cirrhosis) study using simple laboratory tests to predict outcomes in patients with advanced chronic hepatitis C.64

The ability of serum fibrosis markers to predict liver-related mortality was recently assessed in a cohort of 303 HCV-infected patients, 68% of whom were also infected with human immunodeficiency virus (HIV).65 Patients were followed for a mean period of 3.1 years, and an expert committee identified 35 deaths directly or probably attributable to liver disease, with HIV status not affecting the predictability of liver-related mortality. The performance of hyaluronic acid, APRI, and FIB-4 in predicting liver-related mortality was equivalent to that obtained in Model for End-Stage Liver Disease (MELD) or Child-Turcotte-Pugh (CTP) scoring. In predicting 1-year liver-related mortality, the AUROCs for hyaluronic acid, APRI, FIB-4, MELD, and CTP were 0.92, 0.9, 0.9, 0.84, and 0.93, respectively. In multivariate analyses, these markers independently predicted liver mortality in models including MELD or CTP, suggesting that their addition may improve our ability to identify patients at risk of death.

One area of interest is the possible value of these markers for assessing the effects of therapy on liver damage, particularly on liver fibrosis (Fig. 3). Most studies have focused on patients with hepatitis C and B, because there is a specific therapy and a clear definition of response (clearance of viral genome). In this setting, the use of direct fibrogenesis markers and TE appear to detect improvements in patients who achieve a response.66 The lack of a liver biopsy following treatment interruption represents a clear limitation when assessing changes in serum biomarkers. Nevertheless, it has been well documented that a sustained virological response is associated with improvements in both necroinflammatory scores and fibrosis. Recently, Halfon et al.67 studied 114 coinfected patients treated with pegylated interferon and ribavirin who had undergone a liver biopsy both before treatment and 6 months after end of treatment. In virological responders (25%), Forns' score, Fibrotest, FIB-4, Fibrometer, and APRI decreased significantly after viral clearance, correlating with a decrease in the fibrosis stage and in necroinflammatory activity, an outcome not observed in nonresponders.

The combination of fibrosis biomarkers might also be useful for monitoring disease progression or response to lifestyle changes in other contexts, such as reduction in alcohol intake in alcoholic liver disease or weight loss in NAFLD. In the future, fibrosis biomarkers might potentially prove helpful in monitoring the effects of antifibrotic therapies.

Future Steps

Noninvasive assessment of liver fibrosis is already a reality in patients with chronic hepatitis C. In other highly prevalent diseases, such as alcoholic liver disease or NAFLD/NASH, there is still room for improvement, although in the next few years liver biopsy will likely be supplanted in most cases.

The main problems in this area include the lack of a real gold standard to validate the tests and the constant growth in the number of such scores. Although we will have to persevere without a perfect gold standard for the foreseeable future, we clearly need to implement those tests that prove useful. Tests with the highest potential to be widely used are those that: (1) have been extensively validated in independent cohorts of patients, (2) incorporate analytical validation, (3) contain precise information on the diagnostic accuracy and potential causes of unreliable results, and (4) are useful for establishing disease outcomes, which ultimately is the most relevant endpoint.4 In summary, the creation of a categorization system for biomarkers (such as BIPED: Burden of disease, Investigative, Prognostic, Efficacy, and Diagnosis)68 would allow for the more homogenous use of such markers and facilitate their introduction into clinical practice.

Continued research in this area will give us the opportunity to offer our patients more precise and noninvasive diagnostic tools. Liver biopsy will still be part of clinical practice in the coming years, but progress in biomedicine will challenge previously entrenched assumptions and will change our current approach to liver diseases in the next decade. There may yet come a time when liver biopsy will be history.

Ancillary

Advertisement