Statistics from Altmetric.com
- chronic liver disease
- hepatic fibrosis
- liver biopsy
- liver fibrosis
- transient elastography
Liver biopsy and the ‘gold standard’ debate
The birth of Hepatology as a defined clinical discipline coincides with the introduction of Menghini's needle in the late 1950s.1 In those days, very little was known about the aetiology of liver disease and in the majority of patients diagnosis was mainly based on the morphological examination of liver specimens obtained with a biopsy. Features such as necrosis, inflammation and fibrosis were not precisely categorised, and the final label was merely descriptive—that is, chronic persistent, chronic active hepatitis and cirrhosis. In these terms, liver biopsy was the only available diagnostic means, and truly represented a gold standard. The introduction of the so-called ‘one second needle biopsy of the liver’ was characterised since the beginning by a significant cooperative effort between the clinician and the pathologist. Clinical hepatologists were trained in liver pathology and this led to the foundation of the main academic schools of Hepatology in Europe and the USA.
The introduction of scoring systems represented a step forward in the use of liver biopsy. This introduction implied the necessity for more precision in the evaluation of liver biopsy specimens that was beyond the description and, in many instances, the opinion of the pathologist. Overall, the introduction of scoring systems was supposed to facilitate the comparison between samples and the communication of the results according to a global agreement. However, this obvious advantage, once analysed by statistical studies, started to be criticised in term of reproducibility, intraobserver and interobserver agreement, etc. For these and other reasons (sampling error, lack of standards, etc.), the value of liver biopsy as gold standard started to become questionable for clinicians. These issues started to be quite evident when liver biopsy was used to assess the extent of disease progression in terms of fibrotic transformation of the liver tissue, and even more when the fibrotic stage derived from the histopathological evaluation was used as a gold standard to define the diagnostic accuracy of non-invasive methods. Actually, the development of non-invasive means for staging the fibrotic evolution of chronic liver diseases and their clinical validation have highlighted the fact that, overall, liver biopsy is probably an imperfect gold standard.
Indeed, liver biopsy is prone to sampling errors and to intraobserver and interobserver variability.2 3 For instance, even a 25 mm long liver biopsy has a 25% rate of discordance for fibrosis staging.4 Also, when the specimen size is adequate, the level of experience (specialisation, duration and location of practice) of the pathologist may even be more important.5
The performance of any surrogates is classically evaluated by calculation of the area under the receiver operating characteristic curve (AUROC) using liver biopsy as the reference standard. The AUROC represents the probability that a surrogate will correctly rank two randomly chosen patients, one with a liver biopsy considered ‘normal’ and the other ‘diseased’. Because liver biopsy is an imperfect gold standard, a perfect surrogate will never reach the maximal value (1.0).6 Taking into account a range of accuracies of the biopsy and a range of prevalence of significant disease (that influence the AUROC), Mehta et al have shown that in the most favourable scenario, an AUROC >0.90 cannot be achieved even for a perfect marker.6 7
The problem is now how to logically to rely on diagnostic systems for the assessment of hepatic fibrosis in different chronic liver diseases in circumstances in which there is no real gold standard. The aim of this article is to provide for the first time a critical analysis of the current situation based on clinical needs and realistic end points, also considering the introduction of more and more effective treatments.
Clinical needs and end points
Table 1 summarises the current indications for liver biopsy in different fibrogenic chronic liver diseases potentially evolving to cirrhosis. It is quite clear that liver biopsy retains its diagnostic potential in many chronic liver diseases and particularly in those needing a histopathological diagnosis based on fine morphology.
In chronic viral hepatitis due to hepatits B virus (HBV) and hepatitis C virus (HCV) infection, the possibility of efficiently detecting the virus, genotypes and viral load has rapidly decreased the use of liver biopsy as a diagnostic tool. In addition, the progressive increase in the percentage of patients responding to antiviral treatment has made the therapeutic approach more ‘virus based’ than ‘patient based’, and this has reduced the role of biopsy also for the assessment of treatment efficacy. Concomitantly, the value of liver biopsy has been confined to the detection of tissue fibrosis, which is an unquestionable sign of disease progression. This decline in the diagnostic relevance of liver biopsy, particularly in chronic HCV hepatitis, has led to the development of non-invasive methodologies for the assessment or, to use a fashionable term, the ‘prediction’ of liver fibrosis. There are two critical end points—that is, the presence of significant fibrosis which is an indication for antiviral treatment in chronic hepatitis B and C8 9 and the presence of cirrhosis which is an indication for specific monitoring of complications related to portal hypertension and to the increased risk of developing hepatocellular carcinoma. Among the currently available non-invasive methods, there are two different and complementary approaches: (1) a physical approach based on the measurement of liver stiffness using transient elastography (TE)10; and (2) a biological approach based on serum biomarkers of fibrosis11 for which the pioneer has been the FibroTest (a patented algorithm).12 Numerous markers have been proposed since, mainly in hepatitis C, but the most widely used and validated with TE by far are the aspartate-to-platelet ratio index (APRI) (a free non-patented index) and the FibroTest.13–15
For the diagnosis of significant fibrosis, performances of TE and serum biomarkers have been shown to be equivalent in patients with chronic hepatitis C.16 Indeed, TE and several patented indices (FibroTest, Fibrometer17 and Hepascore18) have been recently approved, after an independent systematic review, by the French Health Authorities, for first-line assessment of fibrosis in patients with chronic hepatitis C.19
For the diagnosis of cirrhosis, the situation is different as TE appears to be the most accurate method in patients with HCV when compared with currently available biomarkers and routine blood tests.20
In HBV-infected patients, non-invasive methods are not as well validated as in patients with chronic hepatitis C. Furthermore, more precise fibrosis staging may be required in patients who are candidates for antiviral treatment, since the decision to start treatment and also the type of drugs to be used may be influenced by fibrosis stage. For instance, differentiating between the F1 and F2 stage may be critical in not starting long-term treatment with analogues with the risk of developing resistance. Finally, alanine aminotrasferase (ALT) flares occur frequently in HBV-infected patients and may lead to overestimation of liver stiffness values by TE.21 22 Thus, more data are needed before non-invasive methods can be used routinely in patients with hepatitis B, and liver biopsy should still be recommended before starting antiviral treatment in these patients.8 Conversely, in HBV inactive carriers or immunotolerant patients, in whom liver biopsy is not indicated, non-invasive methods could be useful to strengthen the diagnosis by confirming the absence of significant fibrosis.23 24 In the case of a strong suspicion of cirrhosis, given the excellent performance of TE,25 26 liver biopsy does not seem necessary in order to start antiviral treatment and to screen for complications.
Finally, in order to increase diagnostic accuracy, the combination of TE and biomarkers16 27 28 or biomarkers sequentially29 30 has so far been proposed only in patients with HCV. Such a strategy could lead to a reduction in the number of liver biopsies of >50% for the diagnosis of significant fibrosis and 70% for cirrhosis.31 The advantages of combining two unrelated methods such as TE and biomarkers over the combination of two biomarkers is that TE provides more direct measurement of the liver structure than serum markers and that there is no relationship between the applicability of TE and biomarkers such as the Fibrotest.27
Non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH)
In NAFLD, the second most relevant clinical entity in Hepatology after viral hepatitis, the presence of significant fibrosis does not represent a critical end point in the absence of standardised treatment regimens. However, detection of fibrosis in this clinical setting is highly suggestive of the presence of NASH which may result in cirrhosis. Along these lines, non-invasive methodologies could be employed in patients with NAFLD to select those requiring liver biopsy for a more accurate staging of the disease.
Only a limited number of serum biomarkers (most of them designed for hepatitis C) have been evaluated for their ability to assess liver fibrosis in patients with NAFLD.32–36 It remains difficult, however, to draw any firm conclusions regarding the performance of these markers given the heterogeneous fibrosis scoring systems and end points used in these studies.37 38 The number of studies evaluating TE is also small and they deal with particular populations (children and Asian patients with low body mass index (BMI)).39–41 Given its high negative predictive value and modest positive predictive value, TE could be useful as a screening test to exclude advanced fibrosis in patients with NAFLD.40 Also in this study, the performance of TE for diagnosing severe fibrosis–cirrhosis was better than that of serum biomarkers.
Limitations of non-invasive methods
Table 2 summarises the major advantages and disadvantages of non-invasive methods when compared with liver biopsy. Before analysing the limits of TE and serum markers, it is worth stressing that these two approaches are based on a different rationale and conception: TE measures liver stiffness related to elasticity, which corresponds to a genuine and intrinsic physical property of liver parenchyma, whereas serum biomarkers are combinations of several not strictly liver-specific blood parameters optimised to mimic fibrosis stages as assessed by liver biopsy.6
The use of serum markers directly related to the fibrogenic process, so-called ‘direct markers’ (ie, hyaluronic acid, tissue inhibitor of metalloproteinases 1 (TIMP1), etc.),11 needs carefully to exclude patients with other chronic diseases characterised by fibrogenesis or altered extracellular matrix turnover in other organs and systems. Although the impact of patient age on the performance of direct serum markers has never been thoroughly investigated, the still limited available evidence suggests that the same panel of direct markers performs with higher sensitivity and specificity in a paediatric cohort35 than in an adult cohort34 of patients with NAFLD. Although the applicability and the interlaboratory reproducibility of different tests have been shown to be satisfactory for use in clinical practice,42 43 the interpretation of each test requires a critical analysis in order to avoid false-positive or false-negative results.44
Although TE reproducibility has been shown to be excellent for both interobserver and intraobserver agreement,45 46 its applicability may not be as good as that of biomarkers. Indeed, in our experience, liver stiffness measurements were uninterpretable in nearly one in five cases (failure to obtain any measurement in 4% and unreliable results not meeting the manufacturer's recommendations in 17%).47 The principal reasons were obesity, particularly increased waist circumference, and limited operator experience. These results emphasise the need for adequate operator training and for technological improvements in specific patient populations such as those with NAFLD.
Finally, as the liver is an organ wrapped in a distensible but non-elastic envelope (Glisson's capsula), additional space-occupying tissue abnormalities, such as oedema and inflammation, cholestasis and congestion, may interfere with liver stiffness measurement (LSM), independently of fibrosis. Indeed, as previously mentioned, the extent of necroinflammatory activity has been shown to influence TE measurements in patients with viral hepatitis, with a steady increase of liver stiffness values in parallel with the degree of histological activity.25 45 48 Consistent with these results, the risk of overestimating liver stiffness values has been reported in the case of ALT flares in patients with acute viral hepatitis or chronic hepatitis B,21 22 49 as well as in cases of extrahepatic cholestasis50 or congestive heart failure.51
Could non-invasive tests be used for a better classification of cirrhosis?
While it is increasingly clear that all non-invasive methods perform adequately when used for the identification of advanced fibrosis and cirrhosis, it is still unclear how they could be employed in patients with established cirrhosis. As recently highlighted,52 there is a pressing need for a new classification of cirrhosis integrating histological, clinical, haemodynamic and biological features of this stage of chronic liver disease. This is necessary to overcome the limitation of considering cirrhosis as an end stage of chronic liver diseases,52–54 where the distinction between ‘compensated’ and ‘decompensated’ is mainly defined by clinical outcomes.55 In addition, the cirrhotic stage defined as ‘compensated cirrhosis’ includes anything from the initial histopathological demonstration of ‘early cirrhosis’ to the development of complications of portal hypertension. However, the current tendency to treat patients with cirrhosis with antiviral and, when available, with antifibrotic agents has highlighted the need for classifying compensated cirrhosis by employing parameters more indicative of the fibrotic transformation of liver tissue. In this connection, measurement of the hepatic venous pressure gradient (HVPG) has been proposed for monitoring treatment efficacy in patients with advanced fibrosis and cirrhosis.56 Thus far, the only promising non-invasive approach for monitoring fibrosis progression associated with worsening portal hypertension is LSM obtained by TE. LSM has an excellent correlation with HVPG values below a threshold of 10–12 mm Hg.57 58 In addition, in patients with post-transplant HCV cirrhosis, a similar excellent correlation has been reported between LSM and collagen proportionate area (CPA), a novel approach for the morphometric quantification of liver tissue collagen.59 60 These findings suggest that LSM may be useful for the detection of clinically significant portal hypertension as well as of the extent of collagen deposition within the cirrhotic liver, thereby allowing a subclassification of compensated cirrhosis. On the other hand, LSM may not be accurate in decompensated cirrhosis where, in addition to intrahepatic vascular resistance, there are complex haemodynamic changes.61 Nonetheless, single LSM values or dynamic changes over time could be predictive of decompensation, further decompensation or even death.62 Although retrospective and conducted in a single centre, the study by Foucher et al63 has provided the first ‘proof of concept’ that liver stiffness values may have prognostic value in the context of cirrhosis. Consistent with these results, a recent Japanese prospective study has shown in a large cohort of patients with chronic hepatitis C a correlation between liver stiffness values and the risk of hepatocellular carcinoma.64 Although these findings need to be confirmed in other settings (Caucasian patients or HBV-infected patients) and with longer follow-up, they suggest that TE could be used as a rapid screening tool to allocate patients with cirrhosis to specific risk categories.65
Novel imaging techniques
Alternative imaging techniques including magnetic resonance (MR) elastography, diffusion-weighted MRI, optical digital analysis of CT images of the liver and sonography-based real-time elastography have been proposed.66 The theoretical advantages of these methods include the ability to analyse almost the entire liver and the applicability to patients with obesity or ascites. Preliminary studies in human subjects have confirmed the feasibility of these techniques for quantitative assessment of hepatic fibrosis. For instance, a recent study in 96 patients with chronic liver disease has suggested that MR elastography had a better diagnostic accuracy than TE for the diagnosis of significant fibrosis.67 Although such results are encouraging, these techniques remain so far too expensive and time-consuming for their implementation in clinical practice for screening hepatic fibrosis.
Very recently, a novel imaging technology (acoustic radiation force impulse imaging (ARFI)), involving the mechanical excitation of tissue using short-duration (∼262 μs) acoustic pulses producing propagation of shear waves generating localised micron-scale displacements in the tissue, has been proposed. The shear wave velocity (expressed in metres per second) is measured in a region of interest smaller than that of TE (10 mm long×6 mm wide) but that can be chosen by the examiner. Preliminary results suggest that at least in patients with chronic hepatitis C its performance is very similar to that of TE, although further validation is warranted.68–70 The major advantage of this novel technology is that it can be easily implemented on a modified commercial ultrasound machine. However, as compared with LSM values (2.5–75 kPa), ARFI values are in a very narrow range (0.5–4.4 m/s), which could represent a limitation to its use for making decisions in clinical practice. More data are also awaited regarding the intraobserver and interobserver reproducibility of ARFI.
Need for longitudinal studies
The focus should now shift from cross-sectional diagnosis to utilisation of non-invasive methods in longitudinal studies to look at disease progression, regression and clinical outcomes, and priority should be given to large-scale validation studies. Thus far, very few data are available concerning the performance of non-invasive methods for the evaluation of disease progression or regression in longitudinal studies. For instance, in the context of viral hepatitis these methods could be useful for evaluating fibrosis regression in patients with chronic hepatitis C achieving sustained viral eradication71–73 or for monitoring fibrosis progression in untreated patients. The same could apply in patients with alcoholic liver disease and NAFLD after alcohol withdrawal or improvement of metabolic and clinical parameters.
Screening risk groups and the general population
Most data we have so far regarding liver diseases come from highly selected populations in tertiary care referral centres. Although the exact prevalence of cirrhosis worldwide is currently unknown, it was recently estimated at 76.3 per 100 000 population aged over 25 in the UK (corresponding to 30 000 people) with an increasing incidence between 1993 and 2001.74 Given the high prevalence of undiagnosed cirrhosis in NASH and hepatitis C, these numbers are likely to be an underestimation. Since compensated cirrhosis often goes undetected for extended periods, a reasonable estimate is that up to 1% of populations could have histological cirrhosis.75 Thus chronic liver diseases represent a significant public health problem, with a worldwide mortality attributable to cirrhosis of ∼800 000 deaths per year.76 Obviously tools available so far, such as liver biopsy, are not appropriate for screening fibrosis and cirrhosis in the general population or in at-risk groups such as patients with diabetes or with features of the metabolic syndrome. Given their simplicity and their high acceptability by patients, non-invasive methods could be of value in this particular setting.77 78 Recent studies have suggested that either biomarkers or TE could be used as screening tools to detect liver fibrosis and cirrhosis in the general population.79 80 Along these lines and considering the growing attention being paid to these problems by the medical community, the current tools for the evaluation of the fibrotic evolution of chronic liver disease should be used according to the level of healthcare as illustrated in figure 1. Accordingly, a first screening performed by the general practitioner by employing an accurate physical examination and some easily available tests would allow the selection of patients to be referred to specialised centres for further evaluation.
The experience accumulated in the past few years has further highlighted the fact that there is no perfect method to assess the exact stage of disease progression in fibrogenic chronic liver diseases. Nevertheless, the large intellectual and technical investment made in the development and validation of non-invasive methods has created awareness of a more flexible approach to this clinical challenge. The integration of the information derived from liver biopsy with that of non-invasive tests, TE in particular, allows a more efficient and convenient management of the patient with chronic liver disease. The introduction of the ideal tool—that is, a system allowing the exploration of the whole liver in terms of fibrotic evolution and hepatocellular function—will probably require many years. In the meantime, liver biopsy and non-invasive tools should be employed as an integrated system to maximise their potential, basically acting like two tango dancers.
Competing interests None.
Provenance and peer review Commissioned; not externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.