Biochemical markers of hepatic fibrogenesis: Single measurements are not reliable enough to replace liver biopsy

This is the authors' abstract. We don't add key points for this paper.

See article on page 945 In patients with viral hepatitis and other chronic liver diseases, the stage of fibrosis is the single most important predictor of morbidity and mortality. 1–3 Notwithstanding the problems of sampling error, liver biopsy has remained the gold standard for estimating the extent of fibrosis. On the other hand, liver biopsies are invasive with a small but finite risk of death or serious complications, and are expensive. For several decades, therefore, the search has been on for a simple non-invasive test to reliably estimate the extent of hepatic fibrosis, which would obviate or greatly reduce the need for liver biopsy. Both imaging and biochemical techniques have been studied as surrogate markers of fibrogenesis. Hepatic ultrasound is neither quantitative nor sensitive. Increased echogenicity does not distinguish between fatty infiltration and fibrosis and is therefore non-diagnostic. Similarly, computed tomography (CT) and magnetic resonance imaging (MRI), while useful in the assessment of space-occupying lesions in the liver, have poor sensitivity for the detection of fibrosis especially in its early stages. To be clinically useful, biochemical markers of fibrogenesis should have a high degree of sensitivity and specificity, be easily measurable with good reproducibility, and be relatively inexpensive. Potential markers include the intermediate or final products of extracellular matrix (ECM) synthesis or degradation, or the measurement of levels of enzymes involved in these processes. 4 Markers of functional hepatic mass such as the antipyrine clearance test could be useful but are not in widespread use. 5 To assess the clinical utility of biochemical markers of fibrogenesis, it is necessary to briefly understand its pathophysiology. Chronic liver injury (irrespective of the cause) results in the deposition of a dozen of more ECM molecules that collectively form the hepatic scar. They include the fibril-forming collagens (types I and III), glycoproteins (e.g. fibronectin, laminin, vitronectin) and proteoglycans/glycosaminoglycans (e.g. decorin, heparan, dermatan and chondroitin sulfates, hyaluronan). 6,7 Stellate cells located in the perisinusoidal space of Disse are the major source of ECM. 7 Fibrogenesis is tightly coordinated involving both soluble (cytokines such as TGFβ) and insoluble (the ECM itself) factors secreted in response to liver injury. Fibrosis is not steadily progressive but rather, proceeds in fits and starts depending on the inciting stimulus and the balance of pro- and antifibrogenic factors in the injury milieu. Ultimately, this results in fibrosis with consequent impairment in nutrient exchange, parenchymal and non-parenchymal cellular dysfunction and portal hypertension. Biochemical markers of fibrogenesis include products of collagen synthesis or degradation (e.g. the NH2-terminal propeptide of type III collagen (PIIINP), the 7S domain of type IV collagen (IV-7S), the central helix region of type IV collagen, the carboxy-terminal propeptide of type I procollagen) and serum levels of enzymes involved in matrix biosynthesis (e.g. prolyl hydroxylase, lysyl hydroxylase) or degradation (e.g. tissue inhibitor of metalloproteinase). The ECM glycoproteins (e.g. fibronectin and vitronectin) and proteoglycan/glycosaminoglycans such as serum hyaluronic acid (HA) have also been examined. 4 Serum hyaluronic acid and PIIINP levels have been studied extensively. Estimation of serum markers of fibrogenesis, while conceptually attractive, has limitations. In most instances, markers mirror events at the active interface of matrix deposition; this is characterized by liver-cell injury, inflammation, stellate-cell activation and matrix resorption. Thick scar tissues cause considerable derangement of hepatic function and physiology. Their presence, however, is not reflected in serum as scars are relatively acellular, undergoing minimal turnover. The latter may relate in part to collagen cross-linking, which renders the fibrous tissue resistant to the action of proteinases. Further, the liver is but one of several tissues, including bone and skin, that contain significant amounts of ECM. In health, the liver's contribution to serum levels of intermediates of collagen biosynthesis such as PIIINP is negligible. During fibrogenesis, liver-derived PIIINP levels rise, but remain a small fraction of the total. Clearance of PIIINP is in part by sinusoidal endothelial cells (SEC). 4 Thus, elevated serum levels can be a consequence of compromised hepatic function leading to decreased clearance of the molecule. For these reasons, PIIINP levels must be interpreted with caution during childhood (increased extrahepatic procollagen turnover), acute inflammation (when PIIINP is more a reflection of fibrolysis than fibrogenesis), systemic fibrotic diseases (e.g. rheumatologic disease or pulmonary fibrosis) and in the presence of advanced tumours. Serum HA is currently the best available marker of hepatic fibrogenesis. It is a glycosaminoglycan with a molecular weight of several million Dalton and is synthesized by many tissues, providing elasticity to the ECM. Hyaluronic acid is degraded locally, but a portion enters the systemic circulation via the lymph nodes. Hyaluronan in blood is cleared predominantly by receptor-mediated endocytosis in hepatic sinusoidal endothelial cells, though a small fraction undergoes renal elimination. 8,9 Hepatic production of HA occurs predominantly in the hepatic stellate cell, 10 with levels greatly increased in the fibrotic liver. 11,12 Elevated serum levels of HA in chronic liver disease reflect both increased synthesis by stellate cells and reduced clearance as a result of endothelial cell dysfunction associated with advanced disease. Serum HA levels are increased in the presence of renal disease (impaired clearance), portal hypertension (impaired liver function), glycogen storage disease (reduced SEC function) and diseases associated with increased turnover of ECM (e.g. scleroderma and rheumatoid arthritis). 4 Despite these caveats, serum markers of fibrogenesis have been studied extensively. 13–16 Hyaluronic acid levels correlate more closely with histological fibrosis stage than PIIINP. However, while the correlation coefficients were significant they were only moderate, usually < 0.6. In one study of serum HA levels during α-interferon therapy, levels fell significantly in patients in whom fibrosis improved, increased in those with worsening fibrosis and were unaltered when fibrosis was not modified. 17 In all studies, the range of HA values, even in patients with cirrhosis overlapped that seen in controls. This most likely reflects the variable activity of the fibrogenic stimulus. Hyaluronic acid levels have been examined for the prediction of cirrhosis. The utility of a screening test depends both on the ‘cut-off’ values used based on receiver operating characteristic (ROC) curves as well as the prevalence of the condition. Thus Oberti et al.4 found that a serum HA > 60 μg/L had a sensitivity of 0.97, a specificity of 0.73, a negative predictive value (NPV) of 100% and a positive predictive value (PPV) of only 4% for the prediction of cirrhosis in a General Practitioner population (based on an estimated cirrhosis prevalence of 0.01%). In specialist gastroenterology practice (estimated cirrhosis prevalence 0.3%) the NPV and PPV were 98% and 61%, respectively. In this issue of the journal, McHutchison et al. present data on serum HA levels in a cohort of 486 patients with chronic hepatitis C. 18 Serum HA levels were correlated to histology from the preceding 12 months. Not surprisingly, serum HA levels correlated with the histological stage of fibrosis (R = 0.542, P < 0.001), but also correlated with lobular inflammation and the total histological activity index score. As with the earlier study, 4 these authors found that the main strength of this test resides in its ability to exclude cirrhosis (a high NPV). Using a cut-off value of < 60 μg/L, the absence of cirrhosis was predicted in 99% of cases. The PPV of a level ≥ 60 μg/L for the presence of cirrhosis was 30%. Serum HA levels were also useful in predicting the absence of advanced hepatic fibrosis (stage 3 or 4 fibrosis); NPV 93%, PPV 47%. 18 What then is the clinical value of biochemical markers of fibrogenesis? Clearly, the lack of distinct cut-off values among patients with varying degrees of fibrosis make it unsuitable for estimating fibrosis stage. The high NPV suggests that the main utility of serum HA lies in its ability to exclude cirrhosis. Most strategies for the early detection of complications of chronic liver disease (e.g. varices or primary liver cancer) are based on the diagnosis of advanced fibrosis or cirrhosis. Can the test be used for selecting patients for screening for liver cancer? The high NPV suggests that patients with a low HA level need not be screened. Those with a high level should still undergo a biopsy as the test has a low (47%) PPV for advanced hepatitis C-associated fibrosis. The cost effectiveness of this strategy as the basis of initiating screening in older patients who may not want or are unsuitable for biopsy warrants further study. Can treatment decisions be based on serum HA values? In other words, should only patients with HA levels ≥ 60 μg/L have a liver biopsy, as the test's NPV for the absence of cirrhosis is high? In such a strategy those with cirrhosis on subsequent biopsy will be excluded from therapy as they respond poorly. All others (i.e. those with serum HA < 60 μg/L) and those with values ≥ 60 μg/L and no cirrhosis, would be eligible for combination therapy. Liver biopsy gives an accurate assessment of both the stage and activity of the disease and occasionally alters the diagnosis. Serum HA measurements cannot provide this data. Thus, on current evidence it would appear premature to adopt such a strategy for guiding therapy in patients with chronic hepatitis C. Further, recent evidence suggests that even patients with cirrhosis may sustain a valuable clinical benefit from combination therapy. 19 In summary, single-point estimations of serum hyaluronic acid appear to have relatively little value in the assessment and management of individual patients with chronic liver disease. The data suggest that biochemical indices of fibrogenesis are not yet reliable enough to replace liver biopsy, which should remain the cornerstone of treatment decisions in those with chronic hepatitis C. The main use of fibrosis markers will most likely derive from long-term follow up of patients who have had a baseline biopsy. In these individuals, serum HA estimations may be useful for monitoring progression of the liver disease and to time future biopsies appropriately. Such longitudinal studies looking at the correlation of serum HA levels with hepatic fibrosis over time are urgently needed and awaited with interest.