American Association for Clinical Chemistry
Better health through laboratory medicine
April 2010 Clinical Laboratory News: The Perils of Fatty Liver Disease

CLN Banner Logo

April 2010: Volume 36, Number 4

The Perils of Fatty Liver Disease
Can Biomarkers Replace Biopsy?
By Genna Rollins

Over the past 2 decades, hepatologists have been observing with growing concern a less-publicized consequence of the developed world’s sedentary lifestyle and poor eating habits: fatty livers in people who do not drink alcohol excessively. The conditions associated with this phenomenon, non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH), are the leading causes of chronic liver disease in the U.S. and pose a potential public health crisis. As clinicians grapple with the growing disease burden of NAFLD and NASH, labs may soon be on the frontline of diagnosing, staging, and managing the diseases. Experts predict that panels of serum markers, probably in combination with non-invasive radiologic tests and liver biopsy, eventually will be part of the standard work-up for these conditions. Emerging metabolomic, lipodomic, and genomic information also may one day enhance the diagnostic and disease management picture.

“There's been a change in perception about biomarkers in the field. Until recently, they were regarded as just a flash in the pan, but they are here to stay,” said Neil Guha, MD, PhD, clinical associate professor of hepatology at the University of Nottingham. “It's now a question of when, not if, they'll be in routine use. What we have to do is define what severity of disease can be delineated by current biomarkers.”

A Growing Problem

The disease burden of NAFLD and NASH is pressing, to put it mildly. An estimated one-third of American adults—about 90 million people—have NAFLD, and perhaps as many as one-quarter of those have NASH. Among patients with NASH, nearly one-third develop cirrhosis within 8 years of follow-up. New research also indicates that individuals with NAFLD have increased mortality in comparison to the general population, not only from liver disease but also from cardiovascular conditions and extrahepatic malignancies.

In keeping with these statistics, understanding of the conditions and their potentially serious outcomes has grown with time. “When I finished my fellowship in the early 1980s, we knew fatty liver disease was usually very bad in diabetics and could progress to cirrhosis, but it was a curiosity, something that was discussed at grand rounds,” observed Arthur McCullough, MD, chair of gastroenterology and hepatology at the Cleveland Clinic and professor of medicine at Case Western Reserve University. “In the subsequent 25 years, we’ve come to realize that this is a serious and common disease. And it’s important to realize we’re talking about a spectrum of illness that varies from there not being much to worry about to being a lot to worry about.”

With the prevalence rising and the continuum of fatty liver disease becoming more apparent, clinicians have been crying out for noninvasive tools to effectively diagnose and stage NAFLD and NASH, and monitor patients. Research in this area—especially for serum markers—has been prolific during the past decade, but so far, no single analyte or panel stands out above the others or the admittedly less-than-perfect gold standard, liver biopsy. “The various proposed tests haven’t reached a point to where we have confidence in them as decision-making tools in clinical practice,” explained Jayant Talwalkar, MD, MPH, consultant in gastroenterology and hepatology and associate professor of medicine at the Mayo Clinic. “They haven’t provided an incremental advance in our decision-making in terms of whether or not a patient has fibrosis or needs biopsy. They’ve made some difference, but not enough to get people to change their practice.”

Opinion, however, is divided on this subject. “Nonexpert physicians and patients are waiting for an almost perfect test that is a biomarker with less than 10% of false positive/negative results and more than 99% of applicability. This is not possible, even with liver biopsy,” contended Thierry Poynard, MD, professor of medicine at Group Hospitalier Pitie-Salpetriere in Paris. “It is an illusion to wait for an almost perfect biomarker with adjusted AUROC greater than 90% for the diagnosis of advanced fibrosis” (Adv Clin Chem, 2008;46:131–60). Still, Poynard, who holds the patent for several serum marker panels used in diagnosing liver fibrosis, agrees that further validation of proposed biomarkers specifically in NAFLD and NASH is needed.

A Complicated Disease

The flurry of research has advanced understanding of NAFLD and NASH considerably in recent years. In most patients, fatty liver—simple steatosis—is more or less a benign condition associated with diabetes, obesity, age, sex, ethnicity, dyslipidemia, and the metabolic syndrome. However, as the number of risk factors rise, so does a person’s likelihood of developing NASH, which involves steatosis with hepatocellular inflammation and damage. NASH, in turn, sets-up some patients for progressive fibrosis and cirrhosis and its complications, including liver failure and hepatocellular carcinoma.

NAFLD and NASH seem to develop as a result of a series of insults to the liver. Insulin resistance drives triglyceride production and macrovesicular fat in the liver, through a single factor or the combination of three factors, including decreased hepatic free fatty acid oxidation, raised hepatic de novo lipogenesis, or decreased lipid export from the liver. Once this dysregulated lipid metabolism is in place, it can worsen insulin resistance, creating a vicious cycle. Oxidative stress from agents such as cytochrome P-450 enzymes, endotoxin, cytokines, or environmental toxins promotes lipid peroxidation, which activates pro-inflammatory cytokines like tumor necrosis factor-α, leptin, adiponectin, and C-reactive protein, causing inflammation.

In addition, the white adipose tissue associated with abdominal obesity acts as an endocrine organ, secreting adipokines and cytokines, and furthering inflammation. The tissue injury associated with inflammation changes the metabolism of the extracellular matrix, leading to fibrosis. As the scarring associated with fibrosis progresses, cirrhosis eventually sets in.

More recently, hepatocyte apoptosis has emerged as a major culprit in the progression of NAFLD to NASH. “A number of papers over a short period of time identified that the mitochondria are not functioning properly in people with NASH,” explained McCullough. “We think now that the common final pathway that causes injury in fatty liver disease is due to mitochondrial injury and death. Oxidative stress and inflammation cause mitochondrial injury, and there is some speculation that apoptosis associated with this can stimulate fibrosis itself.”

What's Wrong with Liver Biopsy?

By all accounts, liver biopsy currently is the best diagnostic and staging tool for NAFLD and NASH, but it is far from perfect. Indeed, one author noted wryly that “clearly the liver biopsy standard is not gold but a burnished bronze” (Clin Chem 2004;50:1299–1300). This costly, invasive procedure, which requires patients to take off the better part of a day from work or school, also has risks of complications and death. In addition, liver biopsy is subject to both sampling and interpretation errors, particularly since fibrosis and cirrhosis do not affect the liver uniformly. Complicating matters further, NAFLD and NASH look similar histologically to alcoholic liver disease. With all these caveats, biopsy has an estimated negative predictive value of 74%. The prevalence of NALFD and NASH also presents a practical challenge in terms of the healthcare system not being able to support the volume of biopsies that may be needed in the future.

So as imperfect as liver biopsy may be, for now it remains the gold standard. “It's still an incredibly valuable tool. It tells us not only about the damage in terms of the amount of inflammation, fat and fibrosis, but also about etiology,” explains Guha. “However, using it as a first-line investigation in NAFLD has practical, ethical, and health economic considerations. So what we're trying to find is the balance of when to use liver biopsy. It's not an either-or modality. It depends on the question you're trying to ask.”

The Search for Noninvasive Tools

A variety of non-invasive radiologic markers have been proposed, including ultrasound, computerized tomography, and magnetic resonance imaging, all of which can detect NAFLD but not distinguish NAFLD from NASH. More recently Fibro-Scan, an elastographic device manufactured by Echosens that measures liver stiffness, has shown promise.

Serum markers long have figured into attempts to develop tools that distinguish NASH from NAFLD, accurately stage NASH, and identify which patients with NAFLD or NASH are in the minority who will progress to more advanced disease. Indeed, basic, routine lab tests like alanine aminotransferase (ALT), aspartate aminotransferase (AST), γ-glutamyltransferase (GGT), and platelet count not uncommonly are the first tip-off that there is a problem, since for many people NAFLD and even NASH are asymptomatic. An AST/ALT ratio ≥0.8 has been substantiated as an independent marker of advanced fibrosis in NAFLD, but ALT, which is more specific to altered liver function than AST, has a reported sensitivity of 40%–53% for diagnosing NASH and a specificity of about 50%. Clearly, these tests alone are not good enough. “The problem is, many people who have completely normal AST or ALT levels have significant disease,” noted McCullough.

Use of AST and ALT also has been hampered by standardization issues. According to McCullough, different labs use different normal values and reference ranges for these analytes. At least one study suggested that lower ALT cutoffs of >30 U/L for men and >19 U/L for women may better distinguish NASH, but still with relatively poor sensitivity (Ann Intern Med 2002;137:1–9). The Institute for Reference Materials and Measurements and the International Federation for Clinical Chemistry and Laboratory Medicine (IFCC) issued reference material ERM-AD454/IFCC in 2005 for ALT and reference material ERM-AD457/IFCC in 2009 for AST, and the IFCC Committee on Reference Intervals and Decision Limits is in the process of drafting a publication on reference intervals for AST, ALT, and GGT.

Despite some limitations, these analytes still provide valuable information, and they have been incorporated in several composite prediction models for NAFLD and NASH. For instance, the HAIR score is based on the presence of hypertension, elevated ALT, and insulin resistance index; the BAAT score incorporates body mass index (BMI), ALT, age, and serum triglycerides; and the NAFLD fibrosis score includes age, platelet count, albumin, the AST/ALT ratio, presence of diabetes, and BMI. More recently, the BARD score, which uses BMI, AST/ALT ratio, and diabetes, has been proposed to predict advanced fibrosis in NAFLD patients.

Is a Panel of Biomarkers a Solution?

Without a clear winner among the various testing strategies, researchers have turned to panels of serum markers that reflect the complicated and incompletely understood pathophysiology of NAFLD and NASH. Between 1991 and 2007, researchers proposed at least 14 such panels using between two and seven analytes. Most combine indirect measures of altered hepatic structure and function, such as ALT, AST and platelet count, with direct indicators of oxidative stress, inflammation and fibrosis, the principal pathways of NAFLD-NASH pathophysiology (See Box). Some also include other variables such as age and sex.

NAFLD/NASH Serum Panels

Various serum panels for liver fibrosis have been proposed, but none stand out above the others or in comparison to the gold standard, liver biopsy.



Proprietary (Y/N)

AUROC for Advanced
v. Mild Fibrosis*

(Imbert-Bismut, 2001)

α2M, haptoglobin,
apo-A1, GGT, bilirubin
European Liver
Fibrosis Panel
(Rosenberg, 2004)
FibroSpect II
(Patel, 2004)
α2M, HA,
(Adams, 2005)
α2M, HA, GGT, bilirubin
(Quest Diagnostics)
(Cales, 2005)
α2M, HA, AST, platelets, prothrombin time, urea

(Koda, 2007)

AST, platelets, gamma globulin
(Wai, 2003)
AST/ULN x 100

Legend: apo-A1, apolipoprotein A1; α2M, alpha2-macroglobulin; AST, aspartate aminotransferase; HA, hyaluronic acid; PII-NP, N-terminal propeptide of type III procollagen; GGT, γ-glutamyltransferase;TIMP-1, tissue inhibitors of metalloproteinases

*Fibrosis stage F 2–4 versus F 0–1

Courtesy Dr. Robert P. Myers, University of Calgary, Presentation from the American Association for the Study of Liver Diseases 2008 Liver Meeting

Probably the most investigated of these combination markers is FibroTest, which uses a panel of five analytes—α2-macroglobulin, haptoglobin, apolipoprotein A1, GGT, and bilirubin—and other variables in a proprietary algorithm to provide an estimate of fibrosis stage. FibroTest suffers from some of the same shortcomings as many of the other panels, in that most of the performance data collected thus far has involved chronic hepatitis C rather than NAFLD or NASH, a point the test's developer, Poynard, acknowledges. “FibroTest is less controversial for hepatitis C, hepatitis B and alcoholic liver disease than in metabolic liver disease, because it's difficult to perform a lot of biopsies in that population. The patients have diabetes, metabolic syndrome, and other problems and it's difficult to convince them to have liver biopsies. So of course there is less validation in metabolic liver disease at this time,” he explained.

FibroTest and the other marker panels generally perform well with areas under the receiver operator characteristic curves (AUROC) ranging from 0.75 to 0.86 in differentiating severe fibrosis, but like biopsy they are less discriminating, and typically have low negative-predictive values, for the presence of mild to moderate fibrosis and for discriminating adjacent stages such as mild versus moderate fibrosis. In addition, most of the panels have not been validated longitudinally in sufficiently powered studies.

Some studies also have had methodological issues, according to Anna Mae Diehl, MD, chief of gastroenterology at Duke University School of Medicine. “The way most of these studies have been done is to take a cross-sectional population that differs by stage of fibrosis, perform a liver biopsy, take serum samples, test them, compare them to the biopsy, and then generate positive- and negative-predictive values,” she said. “The problem with this approach is that the serum isn't always sampled at the same time as the liver biopsy, and the liver could have changed in the mean time.” The NIH-funded Nonalcoholic Steatohepatitis Clinical Research Network (NASH-CRN), a consortium of eight clinical centers now in its second funding cycle, has trials in the offing that will address this study design issue. In addition, results from NASH-CRN's two major initial clinical trials, one in adults and one in children, along with various other studies emanating from the first round of funding, are expected to be published this year, she added.

At this point, insufficient data have limited the use of the various proposed NAFLD and NASH biomarker panels. Although two are non-proprietary and others are available as reference tests in the U.S., they have not gained traction in clinical practice. For instance, both Talwalkar and McCullough reported that they only employ the tests for research purposes.

Emerging Diagnostic Tools

Even as the body of evidence grows on the various biomarker panels, new research is leading to still other potential biomarker candidates. Reflecting the latest findings about the role of hepatocyte apoptosis in the pathophysiology of NAFLD, a research team lead by McCullough recently reported that plasma cytokeratin 18 fragment levels correlate with the magnitude of apoptosis and independently predict the presence of NASH, with AUROC estimated to be 0.83 (Hepatology 2009;50:1072–1078). Similarly, an analysis of the lipidomic signature of NASH indicated that although increased lipogenesis, desaturateses, and lipoxygenase metabolites characterize both NAFLD and NASH, impaired peroxisomal polyunsaturated fatty acid metabolism and nonenzymatic oxidation are associated with progression to NASH (Hepatology 2009;50:1827–1838).

In the case of the latter findings, “these provide novel insights into the metabolic basis for activation of inflammation and the disease phenotype in NAFLD. Also the differential changes in these pathways with progression from fatty liver to NASH raises the possibility that these could be used to develop lipodomic signatures to diagnose the disease phenotype,” explained senior author Arun Sanyal, MD, professor of medicine and chairman of gastroenterology at Virginia Commonwealth University Medical Center in Richmond. “It must be emphasized that while these are very encouraging, much additional work is necessary before this can be actualized.”

McCullough agrees that there probably will be a role for genomics and metabolomics in NAFLD and NASH, but with some caveats. “Those kinds of tools are going to be cost-effective if they can help meaningfully increase that approximately 80% AUROC in most of these panels,” he observed. “But an important thing that must be considered too, is the residuals from this work. It may allow us to tailor therapy for NAFLD and NASH so that there won't be one therapy that fits all here. But we're a few years away from that.”

The Role of Labs

As research in the field proceeds, labs not only will have a valuable role in the future, but they also provide vital information with the tools available today. “I would like to emphasize the importance of evaluating global cardiovascular risk in patients with NAFLD,” said Giovanni Targher, MD, researcher in endocrinology in the department of biomedical and surgical sciences at the University of Verona, Italy. “Early and aggressive treatment of underlying cardiovascular risk factors needs to be initiated, as many subjects with NAFLD will have major cardiovascular events and die prior to the development of liver disease.”

Guha emphasized that labs should review the normal ranges for routine lab tests used in the work-up of NAFLD and NASH, as well as participate in validation tests of the various proposed biomarkers. McCullough looks forward to a time when lab and hepatology associations join in developing diagnostic and treatment guidelines for the conditions.

If the science is not settled around NAFLD and NASH, leading researchers remain optimistic that better tools are just around the corner. “We, in the field of NAFLD and NASH, are where the field of viral hepatology was 15 years ago, and look how much progress they've made,” observed Diehl. “At the beginning they also only had a surrogate marker—ALT—which didn't actually measure the hepatitis C virus. So they demonstrated that even without having a great diagnostic test, the tests that are available can be instructive in moving the field forward. Having said that, development of a non-invasive serologic marker was instrumental in getting pharmaceutical companies into the field. Could that happen in NAFLD? I'd say, why not?”

For Further Information:

  • Pagadala M, Zein C, McCullough A. Predictors of steatohepatits and advanced fibrosis in non-alcoholic fatty liver disease. Clin Liver Dis 2009;13:591-606
  • Poynard T, Morra R, Ingiliz P, Imbert-Bismut F, et al. Biomarkers of liver fibrosis. Adv Clin Chem 2008;46:131–60.
  • Wieckowska A, McCullough A, Feldstein A. Noninvasive diagnosis and monitoring of non-alcoholic steatohepatitis: Present and future. Hepatology 2007;46:582–89.
  • Younossi Y. Current management of non-alcoholic fatty liver disease and non-alcoholic steatohepatits. Aliment Pharmacol & Ther 2008;28:2–12.