Applications and Limitations of Noninvasive Methods for Evaluating Hepatic Fibrosis in Patients With Nonalcoholic Fatty Liver Disease

Short abstract

http://aasldpubs.onlinelibrary.wiley.com/hub/journal/10.1002/(ISSN)2046-2484/video/15-4-reading-robinson a video presentation of this article

https://www.wileyhealthlearning.com/Activity/7088585/disclaimerspopup.aspx questions and earn CME

Abbreviations

α2m

α₂‐macroglobulin

AF

advanced fibrosis

ALT

alanine aminotransferase

APRI

AST to Platelet Ratio Index

AST

aspartate aminotransferase

AUROC

area under receiving operator characteristic curve

BARD

BMI, AST/ALT ratio or AAR, and presence of type 2 diabetes mellitus

BMI

body mass index

CI

confidence interval

CLD

chronic liver disease

ELF

Enhanced Liver Fibrosis

FIB‐4

Fibrosis‐4

GGT

γ‐glutamyl‐transpeptidase

MRE

magnetic resonance elastography

NAFLD

nonalcoholic fatty liver disease

NASH

nonalcoholic steatohepatitis

NFS

NAFLD Fibrosis Score

NPV

negative predictive value

PPV

positive predictive value

SWE

shear wave elastography

TE

transient elastography

VCTE

vibration‐controlled transient elastography

Nonalcoholic fatty liver disease (NAFLD) is the most common cause of chronic liver disease (CLD) worldwide and is estimated to affect 25% of the general global adult population.1 Although not all patients with NAFLD, including nonalcoholic steatohepatitis (NASH), progress to development of liver fibrosis, it is important to accurately assess the severity of liver fibrosis to guide monitoring and management decisions given that development of fibrosis is one of the strongest predictors of liver‐related complications and liver‐related mortality (Fig. 1).2, 3 The gold standard for diagnosis of liver fibrosis remains histological assessment via liver biopsy; however, because of the sheer prevalence of this disease and the costs and risks associated with liver biopsy, implementing liver biopsy for all patients with NAFLD is challenging. Liver biopsy is also invasive, costly, and prone to sampling errors.4 Furthermore, the development and availability of several noninvasive methods for the detection and staging of liver fibrosis may be suitable alternatives to liver biopsy in many patients.

Figure 1.

Progression of NAFLD. Adapted with permission from JAMA.3 Copyright 2015, American Medical Association.

Serology‐based scoring systems such as the NAFLD Fibrosis Score (NFS), AST (aspartate aminotransferase) to Platelet Ratio Index (APRI), Fibrosis‐4 (FIB‐4) score, and BARD score use patient information such as age, body mass index (BMI), and standard laboratory values, such as AST, alanine aminotransferase (ALT), platelet count, and albumin, as indices for the evaluation of liver fibrosis.5 Given that NFS, APRI, FIB‐4, and BARD scores are based on standard laboratory values and patient demographics, the ability to calculate and use these fibrosis prediction models are relatively seamless for clinical practice (Table 1). In addition, these scores have a high negative predictive value (NPV) in excluding ≥F3 fibrosis level or advanced fibrosis (AF), and are therefore reasonable screening tools, particularly when more sophisticated tests are unavailable. A recent systematic review and meta‐analysis by Xiao et al.6 compared the performance characteristics of APRI, FIB‐4, BARD, and NFS with liver biopsy. The authors observed that among the four noninvasive fibrosis prediction models, the NFS and FIB‐4 demonstrated the best diagnostic performance for detecting AF when comparing sensitivity, specificity, positive predictive value (PPV), NPV, and area under the receiver operating characteristic curve (AUROC). Although the majority of these serology‐based noninvasive scores were developed for other causes of liver fibrosis, many have since been validated in NAFLD cohorts. For example, McPherson et al.7 evaluated AST/ALT ratio, NFS, and FIB‐4 scores for evaluation of AF in a cohort of 634 patients with NAFLD, and proposed revised cutoffs based on age stratification that demonstrated improved performance characteristics of these scoring systems.

Table 1.

Components, Advantages, and Limitations of Laboratory Evaluation of Fibrosis in NAFLD

Fibrosis Prediction Models	Components	Advantages	Limitations
NFS	Age, BMI, hyperglycemia, platelet count, albumin, aspartate transaminase/alanine transaminase ratio	Low cost	Interpretation of BMI might differ across different ethnic groups
APRI	Aspartate transaminase and platelet count	Low cost	Modest accuracy
FIB‐4	Platelet count, age, aspartate transaminase, alanine transaminase	Low cost	Modest accuracy
BARD	Aspartate transaminase, alanine transaminase, BMI, diabetes	Low cost	Interpretation of BMI may differ across different ethnic groups
ELF test	Hyaluronic acid, tissue inhibitor of metalloproteinase 1, and N‐terminal procollagen III‐peptide	Good prognostic factor for clinical outcomes in patients with CLDs	Age, low CD4+ T cell count, and other factors can affect ELF score results
			Not available for clinical use in the United States
FibroTest (FibroSURE)	GGT, total bilirubin, α2m, apolipoprotein A‐I, haptoglobin	Useful in different CLDs; accurate in patients with overweight or obesity	Suboptimal for early‐stage fibrosis

Data are from Wong et al.5

Other noninvasive fibrosis prediction models include the Enhanced Liver Fibrosis (ELF) panel and FibroTest. The ELF panel incorporates specific fibrosis markers that aim to reflect the degree of fibrosis by assessing extracellular matrix metabolism and consists of hyaluronic acid, tissue inhibitor of metalloproteinase 1, and N‐terminal procollagen III‐peptide. The FibroTest (or FibroSURE in the United States) is composed of α₂‐macroglobulin (α2m), apolipoprotein A1, haptoglobin, total bilirubin, and γ‐glutamyl‐transpeptidase (GGT). Compared with FibroTest, the ELF panel was superior in many characteristics (AUROC: 0.90 versus 0.81, specificity: 90% versus 71%, PPV: 71% versus 31%) when using liver biopsy as the gold standard reference; however, FibroTest demonstrated better performance characteristics compared with ELF when used to rule out AF (NPV: 99% versus 94%, sensitivity: 95% versus 80%) (Table 2).8 Although the FibroTest/FibroSURE is commercially available, the ELF panel is not currently available for clinical practice in the United States. Another noninvasive fibrosis predictor, PRO‐C3 (a marker of type III collagen formation), has emerged as a biomarker for AF in NAFLD. Recently, Daniels et al.9 found that PRO‐C3 is an independent predictor of fibrosis stage in NAFLD and, when combined with age, presence of diabetes, and platelet count (ADAPT score), accurately identifies patients with NAFLD and AF superior to APRI, FIB‐4, and NFS.

Table 2.

Diagnostic Thresholds, AUROC Values, Sensitivities, Specificities, PPV, and NPV of Testing Modalities for Detecting AF in NAFLD

Tests	Cutoffs for AF	AUROC, Mean (95% CI, if Available)	Sensitivity, Mean % (Range, if Available)	Specificity, Mean % (Range, if Available)	PPV, Mean % (Range, if Available)	NPV, Mean % (Range, if Available)
NFS	0.67‐0.67	0.78 (0.75‐0.81)	43.1 (8.3‐100)	88.4 (25.0‐100)	66.9 (26.0‐100)	88.5 (78.6‐100)
APRI	0.54‐0.98	0.75 (0.72‐0.77)	68.6 (61.0‐76.2)	72.7 (59.4‐86)	61.4 (46.9‐76.2)	77.6 (59.4‐94.0)
FIB‐4	1.24‐1.45	0.80 (0.77‐0.84)	77.8 (63.0‐90.0	71.2 (55.5‐88.0	40.3 (24.0‐50.6	92.7 (88.0‐98.0)
BARD	2	0.73 (0.71‐0.75)	75.2 (41.7‐100)	61.6 (32.5‐88.9)	38.3 (15.0‐79.8)	88.7 (49.6‐100)
ELF	0.3576	0.90 (0.84‐0.96)	80	90	71	94
FibroTest (FibroSURE)	0.30	0.81	95.0	71.0	31.0	99.0
VCTE (FibroScan, M Probe)	7.6‐8	0.87 (0.83‐0.90)	87.0 (65.0‐100)	77.2 (65.9‐90.2)	43.4 (27.0‐52.0)	95.5 (86.0‐100)
VCTE (FibroScan, XL Probe)	5.7‐9.3	0.80 (0.78‐0.94)	75.3 (57.0‐91.0)	74.0 (54.0‐90.0)	58.7 (45.0‐71.0)	88.7 (84.0‐93.0)
MRE	3.62‐4.8	0.93 (0.90‐0.97)	85.7 (74.5‐92.2)	98.0 (86.9‐93.3)	71.0 (67.9‐74.5)	93.4 (81.0‐98.1)
2D‐3D SWE	3.02‐10.6	0.91 (0.82‐1.00)	89.9 (88.2‐91.5)	91.8 (90.0‐94.0)	88.2 (83.3‐93.1)	93.4 (92.6‐94.2)

Data are from Xiao et al.6

In addition to serology‐based fibrosis prediction models, multiple imaging methods have become available for the identification and staging of liver fibrosis, such as vibration‐controlled transient elastography (VCTE) or FibroScan, magnetic resonance elastography (MRE), and shear wave elastography (SWE). With the measurement of liver stiffness and controlled attenuation parameter, FibroScan allows for the simultaneous assessment of hepatic steatosis and fibrosis, and is available in two probes, M probe and XL probe, in which the latter is generally used for patients with a BMI ≥30 or with skin‐to‐liver capsule depth of 3.5 to 7.5 cm.10 Recently, Eddowes et al.11 found that only fibrosis stage, not probe type or steatosis, affected liver stiffness measurement when comparing FibroScan examinations (with M and XL probes) and liver biopsy data. The interpretation in patients with ascites and elevated BMI should be made with caution and in light of the overall clinical context given the concern for variability. Furthermore, caution with interpretation of results should also be noted in patients with significant elevation in AST or ALT (e.g., alcoholic hepatitis), which can contribute to overestimation of liver stiffness.

In a systematic review and meta‐analysis, Xiao et al.6 observed that the AUROC of FibroScan M and XL probes for diagnosing AF were significantly higher than those of APRI, FIB‐4, BARD score, and NFS when using liver biopsy as the reference. However, when compared with liver biopsy, the AUROC values for SWE (0.95) and MRE (0.96) were significantly greater than FibroScan M and XL probes, NFS, BARD, FIB‐4, and APRI, with all P values <0.01. In addition, MRE performs accurately in obese patients and those with cirrhosis, whereas SWE and VCTE fall short. However, there are limitations to MRE, including significant cost, time to perform the examination, and lack of availability, because it is not routinely available in all clinical practice settings (Table 3).12

Table 3.

Components, Advantages, and Limitations of Radiological Evaluation of Fibrosis in NAFLD

Tests	Components	Advantages	Limitations
VCTE (FibroScan M and XL Probe)	Mechanically induced impulse	Short processing time (<10 minutes) Ambulatory clinic setting	Requires fasting for 2 hours
	Quantitative measurement of shear wave speed	Immediacy of results	Requires a dedicated device
	Two probes: M and XL (for patients with BMI >30 kg/m2)		Increased variability in patients with ascites or obesity
MRE	Uses a modified phase‐contrast method to image the propagation of the shear wave in the liver parenchyma	Identifies varying degrees of fibrosis in patients with NAFLD	Not readily available
		High concordance with histological severity and percentage collagen area in drug trials	Time‐consuming
		Examination of the whole liver	Costly
		Implemented on a regular magnetic resonance imaging machine
2D‐3D SWE	Ultrasound induced radiation force focus swept over depth faster than shear wave speed to create a Mach cone	Enables simultaneous sonographic imaging of the liver	Requires fasting for 2 hours
	Quantitative measurement of shear wave speed	Implemented on a regular ultrasonography machine	Experienced operators needed
			Quality criteria not well defined
			Increased variability in patients with ascites or obesity

Data are from Wong et al.5

In addition to combining laboratory and imaging data to improve diagnostic accuracy of assessing hepatic fibrosis, other future directions for the detection of fibrosis in NAFLD may lie in genomics. Recent genome‐wide association studies have associated PNPLA‐3 variants and other genetic polymorphisms with AF and steatohepatitis in patients with NAFLD.12 In addition, Loomba et al.13 performed metagenomic sequencing of stool microbiota from 86 individuals with NAFLD and identified 37 species associated with AF, enabling the development of an algorithm that could predict AF with a high degree of accuracy (AUROC 0.93).

Each method for the evaluation of liver fibrosis has its own advantages and limitations. For example, the measurement of liver stiffness in FibroScan may be inaccurate depending on interoperator differences, abdominal adiposity, or size of intercostal space.10 Each test also has its own variability in test performance characteristics, making some tests more well suited for screening in comparison with confirming a diagnosis of AF. Loong et al.14 evaluated the accuracy of combining imaging and serological tests, and found that when using serum test FibroMeter (which consists of α2m, AST, GGT, and prothrombin index) combined with VCTE, the PPV to rule in F2 to F4 and F3 to F4 was increased (PPV: 71.4%‐84.4% for F2‐4 and 61.0%‐88.9% for F3‐4).

The prior study demonstrates how the combination of noninvasive methods for the evaluation of fibrosis may increase the accuracy of diagnosis. This study also speculated that with this high degree of accuracy, liver biopsy could be spared in approximately 50% to 65% of patients. With the sheer volume of patients with NAFLD and the anticipated increase in incidence, the need for accurate assessment of fibrosis is only expected to increase.

In recent years, a wealth of noninvasive assessments of liver fibrosis has been developed, each with its own advantages and limitations. Given the improving accuracy and greater accessibility of noninvasive serology‐based testing for hepatic fibrosis, a reasonable approach is to use them as the first line to “triage” low‐risk patients, who can be safely monitored longitudinally, and distinguish them from high‐risk patients who would benefit from further testing with imaging‐based methods or even liver biopsy. This stepwise method to confirm accurate fibrosis staging may improve the diagnosis and care of patients with NAFLD.