Diagnostic and interventional circulating biomarkers in nonalcoholic steatohepatitis

Abstract Introduction In the setting of the obesity epidemic, nonalcoholic fatty liver disease (NAFLD) has become one of the most prevalent forms of chronic liver disease worldwide. Approximately 25% of adults globally have NAFLD which includes those with NAFL, or simple steatosis, and individuals with nonalcoholic steatohepatitis (NASH) where inflammation, hepatocyte injury and potentially hepatic fibrosis are found in conjunction with steatosis. Individuals with NASH, particularly those with hepatic fibrosis, have higher rates of liver‐related and overall mortality, making this distinction of significant clinical importance. One of the core challenges in current clinical practice is identifying this subset of individuals with NASH without the use of liver biopsy, the gold standard for both diagnostics and staging disease severity. Identifying noninvasive biomarkers, an accurately measured and reproducible parameter, would aide in identifying patients eligible for NASH pharmacotherapy clinical trials and to help tailor intensity of monitoring required. Methods, Results and Conclusions In this review, we highlight both the currently available and novel diagnostic and interventional circulating biomarkers under investigation for NASH, underscoring their accuracy and limitations relevant to our patient population and current clinical practice.


| INTRODUC TI ON
Nonalcoholic fatty liver disease (NAFLD) has become one of the most prevalent forms of chronic liver disease with a global prevalence of approximately 25% among adults. 1 NAFLD is the broad umbrella term that encompasses the spectrum of FLD. Histologically, NAFLD is categorized into nonalcoholic fatty liver (NAFL) or nonalcoholic steatohepatitis (NASH). [2][3][4] To meet diagnostic criteria for NAFL, individuals must have ≥5% hepatic steatosis without evidence of hepatocellular injury. Alternatively, NASH is defined by the presence ≥5% hepatic steatosis with lobular inflammation and hepatocyte injury (ballooning) with or without hepatic fibrosis. 2 It is estimated that approximately 20% of individuals with NAFLD have NASH. 1,2,5 Clinical practice guidelines from both the American and European liver societies currently recommend liver biopsy as the gold standard for diagnosing and staging NASH. 2,6 Enrolment in NASH clinical trials and definition of therapeutic response to novel pharmacologic agents for NASH are also largely defined using histologic criteria. 7 Inclusion criteria for clinical trials generally include fibrosis stage of ≥F2 on biopsy. Primary outcomes assessing response to novel treatment agents are typically defined using changes in the NAFLD Activity Score (NAS) paired with stability or improvement in fibrosis. 7,8 There are several notable limitations in liver biopsy including concerns over sampling error and interrater reliability. 9 In addition, both patients and clinicians are often hesitant to pursue biopsy due to its invasive nature with potential for clinical complications including severe bleeding and rarely death. 10 As a result, liver biopsy is infrequently obtained in clinical practice for diagnosis and staging of NASH. In real-world clinical practice, providers often use a combination of noninvasive serum tests, imaging results and endoscopic findings to arrive at a personalized diagnosis and risk stratification for an individual patient.
The clinical differentiation of NAFL vs NASH is important given the distinct natural disease course for these two subsets of NAFLD.
Individuals with NASH are at risk for developing advanced fibrosis and cirrhosis and therefore have higher overall and liver-related mortality. 2,[11][12][13] NASH patients have also been noted to have significantly higher rates of cardiovascular disease and multiple types of cancer in addition to hepatocellular carcinoma (HCC). 13,14 Recent studies have highlighted the significant clinical implications of fibrosis stage beyond the impact of NASH itself. Individuals noted to have even early stages of fibrosis were found to have significantly increased risk for liver-related morbidity and mortality. [15][16][17] Accordingly, a focus on identifying and monitoring fibrosis stage may have more of a clinical impact than differentiating NAFL from NASH.
Notably, there are heterogeneous rates of disease progression across individuals, making management of NASH challenging. 18 Given that a diagnosis of NASH and fibrosis stage has been clearly linked to risk of clinical outcomes and eligibility for and definition of response to emerging pharmacotherapy, there is a significant unmet need to identify noninvasive diagnostic and interventional circulating biomarkers in NASH. By providing accurate, measurable and reproducible markers to diagnose and monitor NASH activity and fibrosis stage, noninvasive biomarkers will enable us to evaluate risk factors for disease progression and identify patients for pharmacotherapy. Interventional biomarkers are of particular interest as these parameters can assist in monitoring response to treatment. There are multiple significant challenges to identifying accurate diagnostic and interventional circulating NASH biomarkers. These challenges emerge due to the heterogeneous and nonlinear rates of disease progression in NASH and uncertainties in the highest yield parameters for monitoring risk of clinical outcomes. In this article, we summarize the currently available and novel investigative diagnostic and interventional circulating biomarkers in NASH to highlight their current potential role in clinical practice and outline possibilities for future care (Figure 1). F I G U R E 1 Summary of categories of circulating biomarkers in NASH. Overview of the main categories of circulating biomarkers in NASH with summary of specific biomarkers of interest within each category. APRI, AST to platelet ratio index; cfDNA, cell-free circulating DNA; ELF, enhanced liver fibrosis; FIB-4, fibrosis 4; miRNAs, microRNA; NAFLD, nonalcoholic fatty liver disease; PAI, plasminogen activator inhibitor 1; PIIINP, N-terminal type III collagen propeptide; Pro-C3, C-terminal cleavage site of N-terminal type II collagen propeptide; SNP, single nucleotide polymorphism; TIMP1, tissue inhibitor of metalloproteinases 1 | 3 of 12

| A SS E SS MENT OF HEPATI C S TE ATOS IS
In order to meet diagnostic criteria for NAFLD, an individual must have ≥5% steatosis on histology or ≥5.5% intrahepatic triglyceride content by MRI. 2 There are several noninvasive circulating biomarkers that have been assessed to evaluate degree of hepatic steatosis and are outlined below.

| Clinical decision aides
There are several clinical decision aides to assess for hepatic steatosis that combine laboratory data with clinical features ( Table 1). The Fatty Liver Index (FLI) includes triglycerides (TG), gamma-glutamyltransferase (GGT), body mass index (BMI) and waist circumference (WC) and uses ultrasound (US) as the gold standard reference. 19 The FLI has moderate performance characteristics with an area under the receiver operating curve (AUROC) of 0.84, sensitivity (Sn) of 84% and specificity (Sp) of 64%. The Hepatic Steatosis Index (HSI) also uses US as the gold standard reference and is comprised of aspartate aminotransferase (AST)/alanine aminotransferase (ALT), sex, BMI and diabetes mellitus (DM). 20 The HSI has an AUROC 0.81, Sn 93% and Sp 92%. The NAFLD liver fat score uses a more sensitive reference standard, proton magnetic resonance spectroscopy (H-MRS). It is an algorithm that includes fasting serum insulin, AST, AST/ALT ratios, DM and presence of metabolic syndrome (MetS). 21 The NAFLD liver fat score had superior accuracy compared to the FLI and HIS with an AUROC of 0.86-0.87. A decision aide that incorporates more specialized parameters not routinely available in clinical practice is the SteatoTest. This uses the six components of the FibroTest-ActiTest (total bilirubin, GGT, α-macroglobulin, haptoglobin, ALT and apolipoprotein AI), total cholesterol, TG, glucose and BMI adjusted for age and sex. 22 Its diagnostic accuracy is moderate with an AUROC of 0.79-0.80. Lastly, the NAFLD ridge score applies a machine-learning algorithm using laboratory results [ALT, high-density lipoprotein cholesterol (HDL-C), TG, haemoglobin A1c (HbA1c), and white blood cell count (WBC)] with comorbidity data [hypertension (HTN)]. 23 The NAFLD ridge score also uses H-MRS as a gold standard and has very good diagnostic accuracy with an AUROC of 0.87 and a negative predictive value (NPV) of 96%.

| A SS E SS MENT OF NECROINFL AMMATION
The complex underlying pathophysiology of hepatocyte injury involves multiple pathways including but not limited to inflammation, apoptosis, lipid and glucose metabolism and oxidative stress. 24 Given this, it has been extremely challenging to identify noninvasive biomarkers that accurately capture the degree of necroinflammation in NASH. Table 2 outlines the performance characteristics of the most relevant diagnostic and interventional circulating biomarkers for NASH.

| Serum circulating biomarkers of hepatic inflammation
Serum levels of aminotransferases, most commonly ALT, have been frequently applied as routinely available markers of hepatic inflammation in NASH. ALT has consistently been shown to have poor diagnostic accuracy for NASH, with a Sn of 64%, Sp of 75% and an AUROC of approximately 0.60 to detect NASH on liver biopsy in multiple studies. [25][26][27] Researchers are continuously working to identify serum biomarkers that more accurately capture hepatic inflammation in NASH. Plasminogen activator inhibitor-1 (PAI-1) is a serine protease inhibitor that regulates the fibrinolytic system that has been of interest. It has been investigated among patients with biopsy-proven NAFLD and been shown to be associated with underlying NASH. 28

| Adipocytokines
Given that adipocytokines are hypothesized to play a central role in the pathogenesis of NAFL and NASH, these markers have also been the subject of investigation as potential biomarkers for disease severity. Fibroblast growth factor 21 (FGF21) is a hormone-like growth factor involved in several metabolic processes including lipid metabolism and insulin sensitivity. 38

| Circulating biomarkers of oxidative stress
Identifying biomarkers of oxidative stress that correlate with NASH has proven challenging in part due to difficulty in measuring these (HODE) ratio with AST, age and BMI. 44 The oxNASH score provides decent diagnostic accuracy with AUROC ranging from 0.74-0.83, Sn 81% and Sp 97%. 44

| Clinical and biochemical models
Investigators have aimed to improve predictive accuracy by combin-

| A SS E SS MENT OF FIB ROS IS
Investigation regarding noninvasive assessment of fibrosis stage in chronic liver disease has been ongoing for many years and initially was focused among individuals with chronic hepatitis C. More recently, these efforts have shifted to focus specifically on individuals with NASH as these tests have varying accuracy across different disease states. There are a broad array of approaches using circulating biomarkers including clinical decision aides that combine clinical data with serum biomarkers as well as individual markers of extracellular matrix (ECM) turnover (Table 3). Given that fibrosis stage has been strongly associated for risk of clinical outcomes and overall mortality in NAFL and NASH, identifying noninvasive methods to accurately stage fibrosis is essential. 52

| Clinical decision aides
The NAFLD fibrosis score (NFS) is a clinical decision aide computed using platelet count, albumin, AST/ALT and three clinical parameters (age, BMI and glucose intolerance). 53 The NFS has been demon-  57 In clinical practice, approximately 30% of patients will have scores that fall in the indeterminate range for these tests, however, which limits their utility in these instances. 58 There are also limitations in terms of generalizability of the performance characteristics reported in derivation studies to the broader population of patients with NAFLD as these scores were constructed primarily among middle-aged participants who had undergone liver biopsy. 59 The BARD score includes AST/ALT, BMI and DM and generated an AUROC of 0.81 to differentiate patients with NAFL vs those with more advanced fibrosis. 62 Lastly, there is Fibrometer which consists of fasting glucose, AST, ALT, ferritin, platelets, age and weight.
Fibrometer had one of the highest AUROCs to detect significant fibrosis at 0.94. 63 Overall, these noninvasive scoring systems to assess degree of fibrosis are most useful for their NPV, but do have notable limitations in terms of their PPV and thus must be applied correctly to patient care in clinical practice.

| Serum biomarkers of extracellular matrix turnover
There are several panels that incorporate biomarkers of ECM turnover that have been generated to assess correlation with stage of fi-

| Genomics
Accumulating evidence highlights the important interaction between environmental and genetic factors in NAFLD, as reviewed in detail in a recent article by Sookoian et al. 76 MicroRNAs (miRNAs) are short noncoding RNAs that post-transcriptionally regulate gene expression. Their role as biomarkers in NASH is evolving, though present data are insufficient to strongly support their use. miR-122 and miR-34a have been correlated with disease severity in NASH. 77,78 Cell-free DNA (cfDNA) has also been evaluated to assess disease severity in NASH, particularly as it relates to degree of fibrosis. 79 There have been several studies evaluating the role of single nucleotide polymorphisms (SNPs) to evaluate response to lifestyle or pharmacologic interventions in NAFL and NASH. The SNP rs738409 located on GCKR [patatin-like phospholipase domain-containing 3 gene (PNPLA3)] has been identified as a consistent genetic modifier in NAFLD. 80 PNPLA3 I148M variant has been shown to promote hepatic steatosis and stellate cell activation which in turn leads to inflammation and fibrogenesis. 81,82 It has been investigated as a potentially useful biomarker to identify individuals who are more likely to respond to lifestyle interventions or bariatric surgery. 83,84 The rs58542926 polymorphism in TM6SF2 has been associated with reduced hepatic capacity to secrete very low-density lipoprotein and thus has been associated with hepatic steatosis and steatohepatitis. Individuals with the TM6SF2 E167K variant are more susceptible to NASH and appear to have protection against cardiovascular disease. 85,86 The relationship between TM6SF2 rs58542926 polymorphism and risk of NAFLD-related fibrosis is unclear, with studies having conflicting results. The rs780094 polymorphism at the glucokinase regulatory gene (GCKR) locus is also associated with an increased risk of NAFL and in one study among a large cohort of Italian patients was also associated with severity of liver fibrosis. 87,88 A polymorphism in the rs641738 variant of the membrane bound O-acyltransferase domain-containing 7 (MBOAT7) gene, which is involved in phosphatidylinositol remodelling, has been associated with increased hepatic fat content, more severe hepatocyte injury, increased risk of fibrosis and HCC. 89,90 Variation in 17-beta hydroxysteroid dehydrogenase 13 (HSD17B13) which encodes an enzyme localized in lipid droplets within hepatocytes has been associated with protection against hepatic inflammation and fibrosis in the setting of metabolic dysfunction. 91,92 Similarly, a gene variation at the protein phosphatase 1 regulatory subunit 3b (PPP1R3B) is thought to potentially protect against hepatic fat accumulation and decreases risk of progressive liver disease in patients at high risk for NASH. 93,94 Lastly, the rs12979850 polymorphism in the IFNλ3 gene that participates in regulation of innate immunity has been associated with increased hepatic inflammation and fibrosis in patients with NAFLD, particularly in lean NAFLD. 95,96 Several genetic risk scores have been designed to predict the presence of NASH, NASH with fibrosis and NAFLD-related HCC.
These are reviewed in detail elsewhere by Vespasiani-Gentilucci et al 97 A genetic risk score consisting of PNPLA3 rs738409, TMSF2 rs58542926 and Kruppel-like factor 6 (KLF6_rs3750861) was able to identify individuals at risk for NASH cirrhosis among a larger cohort of patients with NAFLD. 98

| Proteomics
Proteomics has been applied to help identify candidate biomarkers in NASH. A group of three priority 1 proteins (complement component C7, insulin-like growth factor acid-labile subunit and transgelin 2) were able to correctly categorize NAFLD patients with NASH with F3/F4 with an AUROC of 0.91. 65

| Lipidomics and metabolomics
It is hypothesized that lipotoxicity resulting from hepatic inflammation is a mediator of hepatic fibrosis progression. 100

| Gut microbiome
Differences in gut microbiome have been evoked in the pathogenesis and risk of disease progression in NASH. It is hypothesized that intestinal microbiota influence hepatic lipid and bile acid metabolism and also contribute to endogenous alcohol consumption. 105 A small study of patients with NAFLD characterized microbiota signatures and noted an increase in Bacteroides among patients with NASH and an increase in Ruminococcus among patients with F2-4 compared to those with no to minimal fibrosis. 106 Interestingly, this is in contrast to findings of another study where there were lower Ruminococcaceae identified among patients with hepatic fibrosis. 107

| SUMMARY
NAFLD is a significant global public health concern given its high prevalence and its associated morbidity and mortality. Emerging data suggest that incorporating novel approaches including genomics, proteomics and the gut microbiome may provide more individualized risk profiles that can better differentiate patients at higher risk of disease progression. Genomics data can potentially be used to assess risk for fibrosis progression and response to therapy and is likely to enter the clinical arena in the future. [76][77][78]83 Proteomics data have shown potential to differentiate NAFL from NASH, whereas lipidomics, metabolomics and the gut microbiome assessments have also been helpful in distinguishing stages of fibrosis in NASH. 65,102,103,108,109 These 'omics' approaches require further validation in larger, more heterogeneous cohorts before they can be considered for use in clinical practice. Ongoing research suggests that combining circulating biomarkers with dynamic imaging modalities may yield better performance than using either modality alone. This combination approach likely represents a mechanism to improve our ability to noninvasively diagnose and monitor patients.

CO N FLI C T O F I NTE R E S T
The author has no conflicts of interest relevant to this manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data used in this review article are available upon request to the author or via open access journal data availability through cited article journal policies.