icon star paper   Hepatitis C Articles (HCV)  
Back grey_arrow_rt.gif
 
 
Noninvasive measures of liver fibrosis- a Review
 
 
  Hepatology Feb 2006
 
Don C. Rockey 1 *, D. Montgomery Bissell 2 1Department of Medicine, and the Division of Digestive and Liver Diseases, University of Texas Southwestern Medical Center, Dallas, TX 2Department of Medicine, and the Division of Gastroenterology and Liver Center, University of California at San Francisco, San Francisco, CA
 
" .....Use of Noninvasive Tools in Clinical Practice
A number of non-invasive tests have been commercialized and are being used more and more in practice. Their use varies widely depending on practice setting and the individual physician's management style. For example, some physicians would treat all patients with HCV, with the possible exception of those with genotype 1 virus and minimal structural disease (fibrosis stage 0/1). For these, a non-invasive test that detects the absence of fibrosis with acceptable accuracy suffices and, moreover, saves the cost and morbidity of liver biopsy. Other physicians (and many patients), weighing the toxicity, limited efficacy, and cost of current interferon-based regimens, would reserve treatment for significant disease (fibrosis stage 2/3) or clear evidence of fibrosis progression. Decision-making in this case requires a test that differentiates minimal disease (stage 0/1 fibrosis) from intermediate fibrosis (stage 2/3). For this purpose, the current generation of non-invasive tests falls short, and liver biopsy still is needed for definitive staging. Fewer clinical data are available in patients with non-HCV chronic liver disease, and use of non-invasive tests is even more variable....."see review of tests below.
 
Abstract
As novel therapies for liver fibrosis evolve, non-invasive measurement of liver fibrosis will be required to help manage patients with chronic liver disease. Although liver biopsy is the current and time-honored gold standard for measurement of liver fibrosis, it is poorly suited to frequent monitoring because of its expense and morbidity, and its accuracy suffers from sampling variation. At the current writing, serum markers and imaging methods are available and increasingly in use as alternatives to biopsy. However, many questions remain about their indications, accuracy, and cost-effectiveness, and more investigation is required before they are put into widespread use. The development of safe, inexpensive, and reliable noninvasive fibrosis measurement tools remains a research priority in clinical hepatology.
 
Article Text
 
Liver fibrosis is the result of chronic injury. Beyond being a marker of injury, it appears to play a direct role in the pathogenesis of hepatocellular dysfunction and portal hypertension. Thus, from a clinical management viewpoint, accurately assessing the extent and progression of fibrosis is important. Liver biopsy is the current gold standard but is poorly suited for active monitoring because of its expense and morbidity. Thus, development of alternatives that are safe, inexpensive, and reliable is a priority.
 
Over the past 2 decades, much has been learned about the mechanisms underlying fibrogenesis and fibrosis, which points to not only potential new therapies but also to tools with which to measure fibrosis. This article reviews the basis for fibrogenesis and the current state of development of fibrosis markers.
 
The Pathophysiology of Fibrosis
 
Definitions
 
Fibrosis.

This refers to a structural change in the liver that accompanies chronic injury. One result may be an increase in liver stiffness, which is the basis for a recently developed non-invasive method of fibrosis assessment using ultrasound. The extracellular matrix (ECM) complexes that constitute fibrosis are not uniform; they differ in age and chemical composition. Over a period of many months, the collagen fibrils of the complex undergo secondary processing, becoming cross-linked. The process confers resistance to degradative enzymes and ultimately irreversibility. For judging the activity of fibrosis, a histological method that distinguishes between old and new fibrosis would be very useful but, as yet, does not exist.
 
Fibrogenesis.
This term denotes production of ECM. It increases in response to injury and is essential to tissue repair. The regulation of fibrogenesis has been studied extensively and is multi-factorial. Inflammation has a central role (see below and Fig. 1). Some components of recently proposed blood tests for fibrosis may detect fibrogenesis rather than fibrosis.
 
Fibrolysis.
The excess ECM produced after injury functions as a temporary scaffolding and is taken down when the repair process is complete. Removal of ECM is mediated by several specific matrix metalloproteinases (MMPs), which in turn are regulated by soluble inhibitors termed TIMPs (tissue inhibitor of metalloproteinase) (Fig. 2). Expression of individual MMPs and TIMPs fluctuates in injury and has been used to monitor fibrosis (or fibrogenesis).
 
Chronic injury of any epithelial tissue elicits a persistent repair response that culminates in fibrosis and scar formation. The response consists of a several-fold increase in collagens and other ECM constituents, including glycoproteins (fibronectin, laminin) and proteoglycans (dermatan sulfate, chondroitin sulfate, and heparan sulfate).[1] The interstitial collagens (types I, III, and V) are particularly prominent during the injury response. A number of ECM components and their breakdown products are being evaluated as candidate markers for fibrosis (Fig. 2).
 
Investigation over the past 2 decades has helped established a cellular mechanism for the wound healing response in liver involving resident stellate cells (also known as lipocytes, Ito cells, or perisinusoidal cells). In the current paradigm, quiescent stellate cells in injury become activated (Figs. 1, 2), undergoing both morphological and functional changes. Morphological changes include loss of vitamin A, acquisition of stress bundles, and development of prominent rough endoplasmic reticulum. Among the functional changes is a striking increase in secretion of extracellular matrix proteins, including collagens, fibronectin, laminin, and proteoglycans, some of which go up more than 50-fold; the extracellular matrix proteins produced by stellate cells closely parallel those found in the whole liver, indicating that this cell type is the principal source of such proteins in injury.
 
The wound repair response represents the integrated response to several kinds of input (Fig. 1). Inflammation is often prominent, and inflammatory mediators are an important stimulus for stellate cell activation and thus fibrogenesis. In viral hepatitis,[2][3] and non-alcoholic steatohepatitis,[4] the available data suggest that injury and inflammation drive activation of stellate cells. Germane to this discussion, many components of these systems are potential fibrosis markers. The ECM itself is an important regulator of stellate cells and the response to injury. Collagens and other ECM proteins engage specific receptors on the cell surface and thereby regulate cell behavior, including transport function and proliferation. The matrix also serves as a reservoir and presenter for soluble factors (cytokines). Injury leads to changes in the composition of the extracellular matrix, which impact these and other cellular functions. These and similar observations provide a biological rationale for linking fibrosis with disease progression.
 
Abbreviations
ECM, extracellular matrix; MMP, matrix metalloproteinases; TIMP, tissue inhibitor of metalloproteinase; AST, aspartate aminotransferase; ALT, alanine aminotransferase; HCV, hepatitis C virus; GGT, gamma glutamyltransferase; APRI, AST to platelet ratio index; MEGX, monoethylglycinexylidide.
 
Fibrosis as a Hallmark of Injury Progression
The liver has a large functional reserve, such that scar buildup is generally inapparent until it becomes extensive and grossly alters the liver structure. An advance of the 1960s was recognizing that the presence of histological fibrosis in an otherwise healthy person is indicative of subclinical disease that in the absence of intervention may progress to liver failure. This led to semi-quantitative histological scoring systems for purposes of staging and providing prognostic information.[5] An activity index also was developed, reflecting the extent of inflammation and hepatocyte necrosis or dropout. Fibrosis with activity carried a graver prognosis than fibrosis or cirrhosis with little or no activity. Over the intervening years, histological scoring systems for both fibrosis and activity have been refined, and variation among trained readers is acceptably low.[6-9]
 
Based on the current paradigm - that injury stimulates fibrogenesis, leading to fibrosis and ultimately liver dysfunction - the evidence would seem compelling that fibrosis per se portends an adverse clinical outcome. However, fibrosis may be static, or it may progress over decades rather than years, and studies of chronic hepatitis C have shown that fibrosis alone may not necessarily be associated with a poor outcome.[10] Further work is needed to correlate fibrosis and clinical outcomes, incorporating nuances such as age of fibrosis, rate of progression, and the nature of the injury factor(s).
 
The Evolving Purpose of Fibrosis Assessment - Why Is Quantitation Important?
 
Forty years ago, liver biopsy was employed fairly frequently, mainly as a diagnostic tool. Its use declined in the 1970s and 1980s as serum markers for hepatitis B and autoimmune diseases (among others) came into vogue. Also, in the absence of effective interventions for most forms of chronic liver disease, biopsy was undertaken mainly to estimate prognosis. However, over the past 10 years, this trend has reversed, largely because of the prominence of chronic hepatitis C. In a study of liver biopsies done in France in 1997, 54% were for evaluation of hepatitis C.[11] Many, if not most, of these were performed to guide the decision to embark on treatment.
 
In hepatitis C, as well as for other diseases (chronic hepatitis B or non-alcoholic fatty liver disease), therapy has improved dramatically but still is applied selectively, for reasons of toxicity and cost, combined with limited efficacy. In general, it is believed that the best candidates for treatment are those with good liver function but evidence of disease progression - i.e., fibrosis - on biopsy. To be useful in patient selection, the ideal test should reliably distinguish a minimum three stages of fibrosis: early/none; intermediate; and advanced/cirrhosis. In the case of chronic hepatitis C, genotype 1, many hepatologists reserve treatment for the group with intermediate fibrosis. In early disease, benefit has not been proved relative to toxicity and cost. In advanced fibrosis, dose-limiting toxicity is a significant problem, and the odds of obtaining a sustained response to treatment are reduced.[12][13] Similar clinical scenarios are evident in other forms of liver disease. Thus, semi-quantitative (at least) staging is important, with liver biopsy being the currently accepted method.
 
An abundance of data now emphasize that fibrosis is dynamic and, with effective intervention, reversible.[14] Successful treatment of viral hepatitis, autoimmune liver disease, alcohol-related disease, schistosomiasis, and others results not only in clinical improvement but also in decreased histological fibrosis.[15-19] As additional therapies are developed, whether for specific diseases or fibrosis per se, fibrosis reduction may emerge as the standard for demonstrating efficacy. Also, in the subset of patients who fail to achieve a response to primary treatment, some patients may exhibit an improvement in liver fibrosis and would potentially be candidates for long-term therapy. A fibrosis test would be extremely useful for confirming that such patients are responding to treatment.
 
Tools Used to Quantitate Fibrosis
 
The ideal noninvasive diagnostic test for hepatic fibrosis is one that is simple, readily available, inexpensive, and accurate. No test meets this definition, although a number of approaches have been proposed and are being evaluated. These include clinical signs, routine laboratory tests, radiological imaging modalities, or quantitative assays of liver function, alone or in combination.
 
Liver Biopsy.
Liver biopsy has been the gold standard for fibrosis assessment because it is direct. However, it is invasive, stressful for patients and their physicians, and is subject to sampling error - even for diseases that appear to affect the liver uniformly (such as viral hepatitis). In a study of 124 patients with chronic HCV infection who underwent laparoscopic-guided biopsy of both the right and left hepatic lobes, in 33% of cases the results were discordant by at least one histological stage (modified Scheuer system).[20] Similar findings have been reported for non-alcoholic steatohepatitis.[21] These findings introduce uncertainty into longitudinal observations involving sequential liver biopsy. Whereas a two-stage change in fibrosis stage is likely significant, a one-stage change may not be. Liver biopsy appears adequate for a 3-level staging of fibrosis but not for a finer delineation. This limits its value, particularly for diseases such as chronic hepatitis C, which typically progresses slowly, on average 0.15 stage per year.[22-25] On the other hand, it appears to be useful for documenting disease resolution, which can be more rapid.[26]
 
Bedside Diagnostic Tools.
The clinical assessment of fibrosis in chronic hepatitis C takes into account the duration of the infection, age at onset, alcohol use, co-existing disease (such as HIV), as well as examination, routine blood tests, and imaging. Its accuracy is reasonable for the ends of the spectrum (no or little fibrosis vs. advanced fibrosis).[27] Few studies have looked systematically at the accuracy of bedside evaluation for other types of liver disease.
 
Cirrhosis may be evident from the signs and symptoms of portal hypertension, and laboratory or radiological data provide clues to subclinical cirrhosis in a substantial number of cases. Relevant literature can be found on the diagnosis of esophageal varices without endoscopy. Various clinical models have been developed.[28][29] Although features such as prothrombin time, albumin level, spleen size, and portal vein diameter (measured by ultrasound) have all been associated with varices, the platelet count has emerged as the best single predictor of esophageal varices and, in some studies, large varices.[30]
 
Routine Laboratory Tests.
Substantial effort has been devoted to investigation of routine laboratory tests for fibrosis assessment (Table 1). In patients with HCV, an aspartate aminotransferase (AST) to alanine aminotransferase (ALT) ratio > 1 has been proposed as a test of cirrhosis.[31][32] Although simple, it has relatively poor sensitivity (53.2%) and negative predictive value (80.7%).[33] Another model in HCV patients included age, gamma glutamyl transferase (GGT), cholesterol, and platelet count[34]; the composite score detected METAVIR stage F2-F4 fibrosis with 94% sensitivity.[34]
 

combined-1.gif

* The tests vary substantially with regard to types of liver disease and numbers of patients studied, although most panels have been studied in patients with HCV. The definition of fibrosis has varied for histological grading systems, as well as in the degree of fibrosis considered to be advanced (for example, some studies consider Metavir stage 2-4 to be advanced, whereas other studies consider Scheuer stage 3-4 to be advanced). Importantly, use of different cutoffs lead to variation in sensitivity/specificity/PPV/NPV (cutoffs recommended by authors were used when possible).
a Studies were retrospective; the specificity for advanced fibrosis ranged from 96%-100% among the 3 studies.
b Also includes age in the panel
c The PGAA added - alpha-2-macroglobulin (Reference [38])
d This panel additionally includes age and history of alcohol intake
e Components tested include: collagen IV, collagen VI, amino terminal
propeptide of type III collagen (PIIINP), matrix metalloproteinase 2 (MMP-2), matrix metalloproteinase 9 (MMP-9), tissue inhibitor of matrix metalloproteinase 1 (TIMP-1), tenascin, laminin, and hyaluronic acid (HA).
Abbreviations: AST, aspartate aminotransferase; GGT, gamma glutamyl transpeptidase; APRI, AST to platelet ratio index; TIMP-1, tissue inhibitor of metalloproteinase 1; ECM, extracellular matrix; HOMA-IR = homeostasis model assessment (for insulin resistance).
 
Another model based on standard laboratory data is the AST to platelet ratio index or APRI; it is the AST level/upper limit of normal divided by the platelet count (109/L) multiplied by 100.[35] The sensitivity and specificity for fibrosis of the APRI value depended on the cut-offs used (i.e., <0.5 to >2.0). For a hypothetical patient with AST 90 IU/L (upper limit of normal, 45) and platelet count 100 (~109/L), the APRI is 2.0, which means the patient has a 41% likelihood of advanced fibrosis and 5% chance of having minimal or no fibrosis. A recent study of 194 patients with chronic hepatitis C found that APRI was superior to the AST/ALT ratio for predicting significant fibrosis but that use of either test circumvented the need for liver biopsy in only a minority of patients.[27] A model that included AST, cholesterol, and insulin resistance (as well as age and an estimate of past alcohol intake) in patients with HCV[36] found that the sensitivity for detection of advanced fibrosis depended greatly on the index value used. At a low probability index, the sensitivity for predicting significant fibrosis was high, but specificity was low, whereas at a high probability index, sensitivity for significant fibrosis was low, but specificity was high; such variation in test characteristics is typical for all of the serum-based tests (and algorithms) currently available.
 
Proprietary Test Panels.
In these panels, selected blood tests and a proprietary algorithm provide an estimate of fibrosis. One of the first was the PGA index, which combines prothrombin time, GGT, and apolipoprotein A1[37] (Table 1). It was later modified to include alpha-2-macroglobulin (the PGAA index).[38] A second-generation panel is the Fibrotest, which combines -2 macroglobulin, haptoglobin, GGT, apolipoprotein A1,and total bilirubin.[39][40] Fibrotest matched biopsy findings in approximately 50% of patients[39] and has been validated in several different cohorts.[41] It was modified by addition of ALT to create a panel for necro-inflammatory activity.[41-43] Its limitations include false-positive results attributable to increases in bilirubin or decreases in haptoglobin; this is particularly relevant to hepatitis C patients on treatment with ribavirin, which commonly causes hemolysis. False-positive results may be caused by Gilbert's syndrome and cholestasis. Acute inflammation causes increases in -2-macroglobulin and haptoglobin, also affecting the test. Finally, although tests of this type are described as detecting fibrosis, they use surrogate markers for fibrogenesis. Although fibrosis and fibrogenesis may correlate, they are not the same thing.
 
Imaging Tests.
Imaging is an attractive approach to fibrosis assessment for 2 reasons. First, radiographic tests are non-invasive, and second, they are better suited to detecting structural changes than are tests of fibrogenesis or inflammation. Routine imaging modalities - ultrasound, computed tomography, and magnetic resonance imaging - are generally capable of detecting advanced disease from the signs of portal hypertension with good sensitivity and specificity. However, they are typically insensitive to mild or moderate fibrosis. Recently, ultrasound technology has been applied to determining tissue elasticity: the method is known as transient elastography, and it uses pulse-echo ultrasound acquisitions to measure liver stiffness. The hypothesis is that fibrosis results in increased stiffness.[44] In a prospective multicenter study of 327 chronic HCV patients by transient elastography, the AUROC[Note a] for METAVIR stage F2-F4 and cirrhosis were 0.79 and 0.97, respectively.[45] In addition, when transient elastography was combined with the Fibrotest, the predictive value for fibrosis stage F2-F4 was improved, with an AUROC of 0.88.[46] The technique is reported to have good reproducibility with low inter- and intra-observer variability. However, it has limitations; the signal penetrates only 25 to 65 mm, which excludes its use in obese patients or those with ascites, it has not been shown to differentiate fibrosis from steatosis, and its sensitivity and precision may not be sufficient for tracking changes in fibrosis as a result of treatment. These concerns notwithstanding, transient elastography is an approach worthy of further study.
 
Specialized Tests of Liver Function.
The initial approaches focused on global tests of liver function. Some detect changes primarily in hepatic perfusion (indocyanine green, sorbitol, and galactose clearance tests); others reflect the metabolic capacity of the liver (the 13C-galactose breath test and 13C-aminopyrine breath test).[47-49] Another, the MEGX test, which measures monoethylglycinexylidide (MEGX) formation after administration of lidocaine, registers the oxidative N-deethylation of lidocaine to MEGX.[50] The sensitivity and specificity of these tests for distinguishing chronic hepatitis from cirrhosis is in the 80% range.[50] They have poor sensitivity for early or intermediate fibrosis.[48][49]
 
Serum ECM Markers of Fibrosis.
As already noted, fibrosis is dynamic, with increased fibrogenesis and fibrolysis, both of which may yield an increased level of circulating ECM components or their fragments. This reasoning has been the basis for several serum tests (Fig. 2), some of which (hyaluronic acid, N-terminal collagen-III propeptide) have been marketed. However, their superiority to standard clinical evaluation is in doubt. For this reason, multi-component tests have been devised, which span a range, from ECM markers to standard laboratory tests, in various combinations.
 
One panel comprising hyaluronic acid, TIMP1, and -2 macroglobulin, termed Fibrospect[51] (Table 1), had an AUROC 0.831 for METAVIR stage F2-F4 fibrosis with positive and negative predictive values of 74.3% and 75.8%, respectively; it did not accurately differentiate specific individual fibrosis stages. Another combination test developed by the European Liver Fibrosis study group examined multiple ECM-related proteins including collagen IV, collagen VI, amino terminal propeptide of type III collagen (PIIINP), and others.[52] Unlike previous studies that focused predominately or exclusively on patients with chronic hepatitis C, this study examined a variety of liver diseases; also, all fibrosis stages were adequately represented (which has been an issue with some studies). An algorithm was developed (and published) that detected moderate or advanced fibrosis (Scheuer stages 3, 4) with a sensitivity of 90%, and the absence of fibrosis (Scheuer stages 0-2) with a negative predictive value of 92%. The AUROC for advanced fibrosis was 0.804. The test appeared to be best in patients with hepatitis C, non-alcoholic fatty liver disease, and alcoholic liver disease.
 
The use of serum tests is being extended through application of proteome technology, which provides a large-scale survey of (up to several hundred) proteins from a single sample. With modifications protein-protein interaction or enzymatic activity can also be examined. Initial studies have suggested that patients with active fibrogenesis will have a distinct fingerprint. In 46 patients with chronic hepatitis B virus, 30 features predictive of significant fibrosis (Ishak stage 3) and cirrhosis were identified. The AUROC for this analysis was 0.906 and 0.921, for advanced fibrosis and cirrhosis, respectively.[53] Another study of 193 patients with chronic HCV was able to distinguish individual METAVIR fibrosis stages with an AUROC of 0.88; this was compared with an AUROC 0.81 for the Fibrotest.[54] A report in patients with HCV correlated differences in several serum proteins with fibrosis[55]; patients with advanced fibrosis had elevated levels of -2-macroglobulin, haptoglobin, and albumin, whereas apolipoprotein A-I, apolipoprotein A-IV, complement C-4, and serum retinol binding proteins were reduced.
 
Through similar methods, a detailed picture of protein glycosylation can be obtained (the glycome). In a study of serum N-glycans,[56] a unique pattern was found in cirrhosis compared with chronic liver disease without cirrhosis. The authors postulated that this may be attributable more to hepatocellular regeneration than to fibrosis per se. When combined with the commercially available Fibrotest, it was 100% specific and 75% sensitive for the diagnosis of compensated cirrhosis.[56] Proteome-based screening is not yet ready for routine use, but refinements can be anticipated. It has potential not only as an improved test of fibrogenesis but also for new insight into the pathobiology of fibrosis.
 
In principle, serum tests reflect the status of the entire liver and therefore may be more accurate than needle biopsy, which is subject to sample variation[20][21] (Table 2). However, available tests distinguish with accuracy only 2 stages of the fibrosis spectrum: F0/F1 (minimal) or F3/F4 (advanced) Intermediate levels of fibrosis are not reliably detected. Additionally, the current tests have not been demonstrated to be useful in tracking changes in fibrosis progression, although they may reflect improvement as a result of treatment.[26] Finally, few of the currently proposed tests have been compared head-to-head or with standard clinical evaluation of fibrosis stage. The published data do not allow comparisons because of the different patient populations examined and the differing methods of analysis.
 

tableComp-2.gif

Use of Noninvasive Tools in Clinical Practice
A number of non-invasive tests have been commercialized and are being used more and more in practice. Their use varies widely depending on practice setting and the individual physician's management style. For example, some physicians would treat all patients with HCV, with the possible exception of those with genotype 1 virus and minimal structural disease (fibrosis stage 0/1). For these, a non-invasive test that detects the absence of fibrosis with acceptable accuracy suffices and, moreover, saves the cost and morbidity of liver biopsy. Other physicians (and many patients), weighing the toxicity, limited efficacy, and cost of current interferon-based regimens, would reserve treatment for significant disease (fibrosis stage 2/3) or clear evidence of fibrosis progression. Decision-making in this case requires a test that differentiates minimal disease (stage 0/1 fibrosis) from intermediate fibrosis (stage 2/3). For this purpose, the current generation of non-invasive tests falls short, and liver biopsy still is needed for definitive staging. Fewer clinical data are available in patients with non-HCV chronic liver disease, and use of non-invasive tests is even more variable.
 
Summary and Future Directions
Our growing understanding of liver injury and fibrogenesis is being used to develop non-invasive tests for fibrosis that are accurate and replace liver biopsy. The current serum tests are a start and may have utility in identifying patients with minimal fibrosis who do not need a liver biopsy. However, as with many new diagnostic technologies, such tests are being adopted and marketed while the evidence of their general utility in various practice settings remains incomplete. For example, there is no convincing evidence that the currently available fibrosis tests have the precision necessary for tracking disease progression in real time or the response to therapy. An important point is that specialized or proprietary tests for fibrosis assessment add to the cost of medical care. Before such tests are accepted, their superiority to routine laboratory studies should be demonstrated. Their use should be restricted to the diseases and questions that have been addressed in well-designed studies. Although debatable given the current body of literature, this is mainly for documenting minimal, or no, fibrosis in patients with chronic hepatitis C.
 
What does the future hold? Refinements of the current generation of non-invasive tests are anticipated. There is room for rational serum-based tests - the recently published European Liver Fibrosis test[52] being an example - that reflect the pathobiology of fibrosis. However, the latter is exceedingly complex, and creating a test that accurately represents the sum of this activity remains a substantial challenge. Proteomic approaches are promising but in their infancy. The new ultrasound-based method, Fibroscan, has the virtue of assessing fibrosis more or less directly and may evolve into a useful and relatively inexpensive test, although its ability to distinguish fibrosis from steatosis remains to be determined. At the technological frontier is nanoimaging,[57][58] which promises to allow direct detection and quantification of both fibrosis and inflammation, potentially circumventing the pitfalls and deficiencies of the surrogates discussed in this review.
 
Endnotes
a The method of analysis for many of these studies uses receiver operating characteristic (ROC) curve analysis, presenting the data as the area under the ROC (AUROC). In this method, 1.0 indicates ideal performance (100% sensitivity and specificity), and 0.5 is no better than chance.
 
 
 
 
  icon paper stack View Older Articles   Back to Top   www.natap.org