Clinical strategy of diagnosing and following patients with nonalcoholic fatty liver disease based on invasive and noninvasive methods

Nonalcoholic fatty liver disease (NAFLD) is an important cause of chronic liver injury in many countries [1, 2]. NAFLD ranges from benign nonalcoholic fatty liver (NAFL) to nonalcoholic steatohepatitis (NASH). The latter condition includes progressive fibrosis, which is associated with high rates of overall and disease-specific mortality [3], hepatocellular carcinoma (HCC) [4, 5] and atherosclerotic diseases [6]. Liver biopsy is recommended as the gold standard method for the diagnosis of NASH and the staging of liver fibrosis in patients with NASH [7]. However, because of its increased cost, risk and healthcare resource use, an invasive liver biopsy is a poorly suited diagnostic test for such a prevalent condition [8]. Therefore, the development of reliable noninvasive methods for the assessment of liver fibrosis has become essential to estimate the potential progression of NASH to cirrhosis and HCC and to guide therapy. In this review, we highlight recent advances in biomarkers, scoring systems, ultrasound (US)-based elastography techniques and magnetic resonance imaging (MRI) techniques with which to evaluate the liver fibrosis stage and steatosis grade and discuss their usefulness for surveillance of the liver status, including the presence or absence of HCC, in patients with NAFLD.

Histological analysis of liver biopsy samples has played a central role in the management of NAFLD and NASH in terms of diagnosis, definition of severity, and prediction of prognosis. Hepatic steatosis is a histological hallmark of NAFLD. Hepatic steatosis independent from alcohol consumption in morbid obesity and diabetes had been reported in the decades after 1952 [9‐11] and Ludwig et al. finally propounded NASH in 1980 [12]. Studies in the 1970s identified that most fatty liver did not progress to fibrosis and cirrhosis [13, 14]. Therefore, hepatic steastosis was considered as a benign or non-harmful findings; however, it was unknown whether the other concomitant findings including lobular inflammation, Mallory–Denk bodies, ballooning degeneration and fibrosis were associated with the disease progression of NAFLD/NASH.

In 1999, Matteoni et al. confirmed the progressive course of patients who had NAFLD with ballooning degeneration [15]. As a result of these differences, these authors classified NAFLD with hepatic steatosis and ballooning degeneration as type 3 and NAFLD with liver steatosis, ballooning degeneration, fibrosis, and Mallory–Denk bodies as type 4 (Table 1). Types 3 and 4 NAFLD are associated with higher liver-related mortality rates than type 1 (only liver steatosis) and type 2 (liver steatosis and lobular inflammation) [15]. NAFLD can be divided into NAFL without ballooning degeneration and NASH with ballooning degeneration [16, 17]. Matteoni’s classification, which is based on analysis of the liver-related prognosis, has been used as the standard classification system for the diagnosis of NASH. However, this study and classification by Matteoni et al. left three main issues for other researchers to resolve over the next decade: a too-small sample size of patients with type 2 and 3 NAFLD (n = 10 and 19, respectively) to confirm the impact of ballooning degeneration on prognosis, intraobserver and interobserver differences in the diagnosis of ballooning degeneration, and a missing classification for the disease severity of NAFLD [18, 19]. The classification published by Brunt et al. in 1999 enabled definition of the severity of NASH as an activity grade (grades 1–3) and fibrosis stage (stages 1–4) [20] (Table 1). The NAFLD activity score (NAS) was developed by the NASH Clinical Research Network Pathology Committee based on Brunt’s classification and is mainly used to judge treatment responses or disease progression in clinical studies [21] (Table 1).

The modified diagnostic criteria for NASH published by Younossi et al. in 2011 were a turning point in the diagnosis of NASH (Table 1). In their studies, NASH was diagnosed in the presence of: (1) any degree of steatosis along with centrilobular ballooning and/or Mallory–Denk bodies or (2) any degree of steatosis along with centrilobular pericellular/perisinusoidal fibrosis or bridging fibrosis in the absence of another identifiable cause [22, 23]. The most important difference from Matteoni’s classification is the significance of ballooning degeneration and fibrosis; according to the modified criteria, NASH can be diagnosed without ballooning degeneration if fibrosis is identified. Younossi et al. also performed a multivariate analysis of prognosis, and liver fibrosis was the only histological finding independently associated with liver-related mortality [23]. This result confirmed that liver fibrosis in patients with NASH is more important than ballooning degeneration in terms of the liver-related prognosis of NASH. In a multi-center study performed by Angulo et al. [24] in 2015, the severity of liver fibrosis was correlated with the hazard ratio of the overall morality or receipt of liver transplantation (hazard ratio of 1.92 in patients with stage 3 fibrosis and 6.35 in those with stage 4 fibrosis compared with patients without fibrosis). Dulai et al. recently performed a meta-analysis with similar results with respect to an overall mortality rate that increased with progression of the fibrosis stage [25]. In terms of liver-related mortality, they also found that the mortality rate ratio increased exponentially as the fibrosis stage increased compared with patients without fibrosis (1.4 in patients with stage 1 fibrosis, 9.6 in stage 2, 16.7 in stage 3, and 42.3 in stage 4) [25].

Consequently, liver fibrosis could be considered the most clinically important histological finding of NAFLD. Another advantage of focusing on liver fibrosis is its high reliability. In addition, the widespread general use of noninvasive alternatives to histological evaluation of liver fibrosis, including vibration-controlled transient elastography (VCTE™, FibroScan; Echosens, Paris, France) and blood biomarkers, might paradoxically prove the importance of histological liver fibrosis again in future clinical studies.

Efforts to diagnose NASH or non-NASH using histological findings other than liver steatosis and fibrosis now seem to be less significant in the context of long-term clinical outcomes; however, whether improvement of liver fibrosis can contribute to prolonged survival and decreased liver mortality remains unknown. On the other side, the latest updated practice guidance from the American Association for the Study of Liver Diseases (AASLD) reconfirmed the importance of ballooning degeneration, lobular inflammation, and Mallory–Denk bodies for diagnosing NASH [26]. The histological findings that are associated with better outcomes should be confirmed because they might represent a treatment target in the era of developing new agents for NAFLD.

As previously stated, liver biopsy is defined the gold standard for the diagnosis of NASH. However, the pathological diagnosis of NASH has several limitations such as interobserver variability, absence of diagnostic consensus, and sampling error. Various noninvasive biomarkers have been proposed to detect NASH for avoiding redundant liver biopsies. These biomarkers include indicators of insulin resistance (homeostatic model assessment-insulin resistance [HOMA-IR], adiponectin), oxidative stress (thioredoxin, advanced glycation end products [AGEs]), inflammation (high-sensitivity C-reactive protein, tumor necrosis factor-α), and apoptosis (cytokeratin-18 fragment) as well as hormones (insulin-like growth factor 1, dehydroepiandrosterone sulfate, and free testosterone) and hepatic fibrosis markers (Fig. 1). Platelet (PLT) count is the simplest index for predicting advanced fibrosis in patients with NAFLD [27]. Notably, PLT levels may be unexpectedly high even when hepatic fibrosis is advanced (18.9 × 10⁴/μL in patients with stage 3 NASH). Fibrosis markers also have been extensively examined, including hyaluronic acid, type IV collagen 7S, procollagen III peptide and Wisteria floribunda agglutinin-positive Mac-2-binding protein (WFA+ -M2BP). A recent meta-analysis showed that measurement of the serum level of WFA+ -M2BP, which is now covered by the health insurance system in Japan, is useful for detecting severe fibrosis or predicting HCC development in patients with chronic liver disease [28]. WFA+ -M2BP is also of clinical use for assessing liver fibrosis in patients with NAFLD [29, 30]. Plasma collagen type III, which is a neo-epitope marker that reflects true type III collagen formation [31], is a useful test with which to predict fibrogenesis and monitor disease progression [32]. Recently, Kamada et al. demonstrated that serum levels of fucosylated haptoglobin can reflect histological hepatocyte ballooning in NAFLD-affected liver [33, 34]. However, these fibrosis markers have some drawbacks such as high cost, difficulty of generalizability in clinical practice, and lack of sufficient evidence among different ethnicities.

Numerous non-invasive panels of tests have been developed for the staging of liver disease consisting of combinations of clinical and routine laboratory parameters, as well as specialized tests. The aspartate aminotransferase-to-alanine aminotransferase ratio (AAR) is also a simple index with which to detect advanced fibrosis, although it has limitations when used alone [35]. It is accepted that patients with NAFLD who have an AAR of > 0.8 are likely to have advanced fibrosis, as observed in the BARD score (which comprises a body mass index [BMI] of > 28 kg/m² [= point], AAR of > 0.8, and the presence of diabetes) [36]. The NAFLD fibrosis score (NFS) [37] and the fibrosis-4 (FIB-4) [38] index have been externally validated in populations of different ethnicities with consistent results [39]. These scores have been proven useful for not only excluding advanced fibrosis (stage ≥ 3) but also predicting liver-related mortality [40, 41]. The combination of the PLT count (< 15.3 × 10⁴/μL), albumin concentration (< 4.0 g/dL), and AAR (> 0.9; i.e., the PLALA score) is a good predictor of cirrhosis in patients with NAFLD based on data from the Japan Study Group of NAFLD (JSG-NAFLD) group [42]. Surprisingly, the cut-off serum albumin level was higher than expected. The serum albumin level should be monitored to avoid misdiagnosis of cirrhosis in patients with NAFLD. A multicenter study by the JSG-NAFLD showed that measurement of the serum ferritin, fasting insulin, and type IV collagen 7S, the combination of which is called the NAFIC score, is the most useful method with which to detect NASH according to the data of 619 Japanese patients with biopsy-proven NAFLD [43]. Okanoue et al. recently suggested that the combination of type IV collagen 7S and aspartate aminotransferase (CA index-NASH = 0.994 × type IV collagen 7S + 0.0255 × aspartate aminotransferase) is a reliable and simple scoring system with which to differentiate NASH from NAFLD. Patients with NAFLD with a CA index-NASH of > 7 are likely to have NASH [44]. These scoring systems from Japan (NAFIC score and CA index) should be validated in an independent population.

Genetic factors are important for the development of NAFLD, as well as environmental factors. Recent genome-wide association studies revealed that genetic variation, rs738409 (I148M), in the patatin-like phospholipase domain-containing protein 3 (PNPLA3) on chromosome 22 influences NAFLD [45]. It is associated with NASH also in Japanese patients with NAFLD [46, 47]. In one study, the prevalence of the PNPLA genotypes GG, CG, and CC was 20, 50, and 30%, respectively, in the Japanese general population [48]. Interestingly, the PNPLA3 G allele is more prevalent in Japanese than Caucasian individuals [48, 49]. In addition, PNPLA3 GG homozygosity is strongly associated with hepatocarcinogenesis in patients with NAFLD [49, 50].

Transmembrane 6 superfamily 2 (TM6SF2) has broad tissue and organ expression with highest relative levels of expression in the small intestine and liver [51]. The chromosomal location of the TM6SF2 gene in human is 19p13.11. One variant in TM6SF2 (rs58542926, E167K) is positively associated with elevations in serum AST and ALT [52]. Exome-wide association study (EWAS) identified a variant of this gene that is associated with NAFLD [52]. The TM6SF2 rs58542926 variant was strongly associated with NAFD, advanced fibrosis, and cirrhosis, independent of age, BMI, type 2 DM and PNPLA3 [53]. Both the PNPLA3 and the TM6SF2 have been examined in recent phase 2 and phase 3 clinical trials. In the near future, it is expected that determination of PNPLA3 and TM6SF2 single-nucleotide polymorphisms will be available in daily clinical practice.

Conventional US imaging plays a major role in the diagnosis of chronic liver disease and is routinely used for this purpose. Furthermore, US is the most common technique for evaluation of hepatic steatosis because of its availability, lack of radiation exposure, and low cost. The mean sensitivity of US for identification of steatosis ranges from 73.3 to 90.5% [54]. The sensitivity of US for detecting mild steatosis (0–10% on liver biopsy) is relatively lower. A decreased steatosis grade is observed with progression to advanced liver fibrosis, especially cirrhosis in patients with NAFLD [55]. In addition, conventional US cannot accurately differentiate the various stages of liver fibrosis in patients with NAFLD. Therefore, conventional US is not able to discriminate among steatosis, fibrosis, inflammation, and existence of NASH [56, 57] (Supplementary Table 1). To assess liver fibrosis, several noninvasive US-based elastography techniques have been developed. These methods include VCTE, acoustic radiation force impulse (ARFI) imaging and shear wave elastography (SWE). We herein discuss the principles, advantages, diagnostic accuracy, and limitations among different US-based elastography techniques for predicting liver fibrosis in patients with NAFLD.

Magnetic resonance elastography (MRE) is an MRI-based method for quantitative imaging of tissue stiffness. It is available from several manufacturers of MRI scanners as an option that includes special hardware and software. Quantitative stiffness images (elastograms) of the liver can be rapidly obtained during breath-hold acquisitions and can therefore be readily included in conventional liver MRI protocols [91]. This is performed using similar principles to VCTE, whereby propagating shear waves are generated and imaged using phase-contrast MRI, which includes oscillating motion-sensitizing gradients. Inter-observer agreement for liver fibrosis staging in patients with viral hepatitis B and C is reportedly almost perfect for both histopathology (ICC = 0.91; 95% confidence interval [CI], 0.86–0.94) and MRE (ICC = 0.99; 95% CI, 0.98–1.00), with significantly higher agreement for MRE [92].

The surveillance system for HCC in patients with NAFLD is problematic. Even in Japan, the estimated number of patients with NAFLD is around 10–20 million. In one study, the incidence of HCC among patients with NAFLD was 0.44 per 1000 person-years (range, 0.29–0.66), and that in NASH was 5.29 per 1000 person-years (range, 0.75–37.56) [107]. Thus, the incidence of HCC is considerably low, and it is impossible and uneconomical to perform imaging examinations for HCC surveillance in all patients with NAFLD.

The risk factors for HCC in patients with NAFLD are important to surveillance systems. The most important risk factor of developing HCC is advanced fibrosis (fibrosis stage 3 and 4). PLT count of < 19.2 × 10⁴/µL and serum hyaluronic acid level of > 42 ng/mL are considered to be good predictive markers of advanced fibrosis [27, 108]. As previously stated, several reports have recommended liver stiffness of > 9 kPa as measured by the VCTE as the cut-off value for grade 3 fibrosis in patients with NAFLD [109]. Therefore, if the PLT count is < 19.0 × 10⁴/µL, hyaluronic acid level is > 40 ng/mL, or liver stiffness is > 9 kPa as measured by the VCTE, advanced fibrosis is suspected. In addition, Kawamura et al. reported that the risk factors for HCC were advanced fibrosis, male sex, old age, and diabetes [110]. Kodama and Tokushige reported a high gamma-glutamyl transferase level in addition to the above-mentioned factors [111, 112]. Ashca et al. reported past or social drinking as a risk factor [113]. Such patients should undergo US and/or other examinations every 6 months because these patients have an elevated risk of both HCC and cardiovascular events [114].

About 85% of NAFLD-HCC developed from NAFLD with advanced fibrosis (grade > 3) in female patients with NAFLD-HCC, whereas about 60% of HCC developed from NAFLD with advanced fibrosis in male patients. In general, patients with mild fibrosis were older age and had poor control of diabetes in hepatitis C [115]. Therefore, male patients with NAFLD who are older and have poorly controlled diabetes should be taken for adequate surveillance of HCC, even if their hepatic fibrosis is not advanced. At the present time, surveillance algorithms for HCC have already been described in guidelines published by the AASLD) the European Association for the Study of the Liver and the European Organization for Research and Treatment of Cancer (EASL-EORTC) and the Japan Society of Hepatology (JSH). The JSH algorithm considered the functional imaging techniques of gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid-enhanced MRI (EOB-MRI) and Sonazoid contrast-enhanced ultrasound (CEUS) to be very important diagnostic modalities, bet the AASLD and EASL-EORTC algorithms suggest that a diagnostic be made based solely on hemodynamic findings using dynamic CT/MRI and biopsy findings [116].

Staging of liver fibrosis is essential in determining the prognosis and optimal treatment for patients with NAFLD and in guiding surveillance for the development of HCC. Several epidemiological studies have revealed that cardiovascular disease is the most common cause of death in patients with NAFLD independent from other metabolic comorbidities and that liver-related mortality is the second or third most common cause of death [24, 117‐119]. Notably, histological liver fibrosis is the only independent predictive factor for the overall long-term outcome of NAFLD, including cardiovascular disease and liver-related mortality. Liver biopsy is recommended as the gold standard method for the diagnosis and staging of fibrosis in patients with NAFLD as well as other chronic liver diseases [120]. However, this procedure is associated with a risk of complications, and is costly and time-consuming for both providers and patients [8]. Furthermore, it is impossible to enforce liver biopsy in all patients with NAFLD because the estimated number of such patients has reached 80–100 million in the USA and 10–20 million or even more in Japan [121]. In this review, we have introduced several noninvasive diagnostic methods for NAFLD with a particular focus on liver fibrosis and steatosis as well as liver biopsy.

Progression of chronic liver disease is often asymptomatic; however, patients usually present with complications at advanced stages of the disease. Noninvasive scoring systems are considered useful for selection of patients who require further inspection of liver fibrosis. US-based elastography and MRE are now becoming more widespread because they have demonstrated higher accuracy in the diagnosis of severe fibrosis and liver cirrhosis. The major advantages of these elastography techniques compared with liver biopsy are that they are painless and rapid, have no associated complications, and are universally accepted by patients. Additionally, ARFI elastography, SWE, and MRE can be performed during standard examinations of the liver that are routinely performed for HCC surveillance in patients with chronic liver disease (Table 2). Even when the diagnosis of cirrhosis is obvious, there are two additional benefits of LSM in these patients. First, a higher degree of stiffness in patients with cirrhosis could be of diagnostic value for detecting the presence of large varices [122]. Second, greater stiffness may be predictive of other complications and subsequent liver-related death [123]. In this regard, we proposed a clinical algorithm for diagnosing and following the patients with NAFLD based on liver biopsy and noninvasive methods such as scoring systems, US-based elastography and MRE, and appropriate surveillance of HCC (Fig. 2).