EASL Clinical Practice Guidelines


Currently available non-invasive methods

Non-invasive methods rely on two different approaches: a “biological” approach based on the quantification of biomarkers in serum samples or a “physical” approach based on the measurement of liver stiffness (LS). Although these approaches are complementary, they are based on different rationales. Serum biomarkers indicate several, not strictly liver specific clinical and serum parameters that have been associated with fibrosis stage, as assessed by liver biopsy, whereas LS corresponds to a genuine and intrinsic physical property of liver parenchyma.

Serum biomarkers of liver fibrosis

Many serum biomarkers have been proposed for staging liver fibrosis, mainly in patients with chronic hepatitis C. They are summarized in Table 2. The FibroTest® (proprietary formula; Biopredictive, Paris, France, licensed under the name of Fibrosure® in the USA (LabCorp, Burlington, NC, USA)) was the first algorithm combining several parameters [21]. Several other scores or algorithms have been proposed in hepatitis C virus (HCV) [[22], [23], [24], [25], [26], [27], [28], [29], [30], [31],[32], [33], [34], [35]], as well as in hepatitis B virus (HBV) [[36], [37]], human immunodeficiency virus (HIV)-HCV coinfection [[38], [39]], and NAFLD [[40], [41]]. Four are protected by patents and commercially available: the FibroMeter® (Echosens, Paris, France), the FibroSpectII® (Prometheus Laboratory Inc. San Diego, CA, USA), the ELF® (Enhanced Liver Fibrosis Test, Siemens Healthcare, Erlangen, Germany) and the HepaScore® (PathWest, University of Western Australia, Australia). Non-patented methods use published models, based on routinely available laboratory values.

Table 2
Currently available serum biomarkers for non-invasive evaluation of liver fibrosis in chronic liver disease.

∗Graded as 0–2.

The practical advantages of analyzing serum biomarkers to measure fibrosis include their high applicability (>95%) [42], their good inter-laboratory reproducibility [[43], [44]], and their potential widespread availability (non-patented) (Table 3). However, none are liver specific and their results may be influenced by changes in clearance and excretion of each individual parameters. For instance, increased levels of hyaluronate occur in the post-prandial state [45] or in aged patients with chronic inflammatory processes such as rheumatoid arthritis [46]. Also, the reproducibility of measurement of some parameters included in “indirect" serum markers, such as aspartate aminotransferase (AST) levels or platelet count, is questionable [47]. In addition, the interpretation of each test requires a critical analysis in order to avoid false positive or false negative results. For instance, when using FibroTest®, the existence of hemolysis or Gilbert syndrome that can lead to false positive results (by a decrease haptoglobin or an increase in bilirubin, respectively) should be taken into account [48]. Similarly, acute hepatitis can produce false positive results in the aspartate-to-platelet ratio index (APRI), Forns index, FIB-4 or FibroMeter® tests, since all include serum levels of aminotransferases in their formulas.

Table 3
Respective advantages and disadvantages of currently available non-invasive methods in patients with chronic liver disease.

ROI, region of interest.

Liver stiffness measurement

Transient elastography

Liver fibrosis can be staged using 1-dimensional ultrasound TE (FibroScan(R), Echosens, Paris, France) [49], which measures the velocity of a low-frequency (50 Hz) elastic shear wave propagating through the liver. This velocity is directly related to tissue stiffness, called the elastic modulus (expressed as E = 3 ρv2, where v is the shear velocity and ρ is the density of tissue, assumed to be constant). The stiffer the tissue, the faster the shear wave propagates.

TE is performed on a patient lying supine, with the right arm elevated to facilitate access to the right liver lobe. The tip of the probe is contacted to the intercostal skin with coupling gel in the 9th to 11th intercostal space at the level where a liver biopsy would be performed. The operator, assisted by a time-motion image, locates a liver portion at least 6 cm deep and free of large vascular structures. The operator then presses the probe button to start the measurements (“shots"). TE measures LS in a volume that approximates a cylinder 1 cm wide and 4 cm long, between 25 mm and 65 mm below the skin surface. The software determines whether each measurement is successful or not. When a shot is unsuccessful, the machine does not return a value. The entire procedure is considered to have failed when no value is obtained after ten shots. The final result of a TE session can be regarded as valid if the following criteria are fulfilled: 1) a number of valid shots of at least 10; 2) a success rate (the ratio of valid shots to the total number of shots) above 60%; and 3) an interquartile range (IQR, reflecting the variability of measurements) less than 30% of the median LS measurements (M) value (IQR/M ⩽0.30%) [50].

The results are expressed in kilopascals (kPa), and range from 1.5 to 75 kPa with normal values around 5 kPa, higher in men and in patients with low or high body mass index (BMI) (U-shaped distribution) [[51], [52], [53], [54]].

Advantages of TE include a short procedure time (<5 min), immediate results, and the ability to perform the test at the bedside or in an outpatient clinic (Table 3). Finally, it is not a difficult procedure to learn which can be performed by a nurse or a technician after minimal training (about 100 examinations) [55]. Nevertheless, the clinical interpretation of TE results should always be in the hands of an expert clinician and should be made with full knowledge of patient demographics, disease etiology and essential laboratory parameters.

Although TE analysis has excellent inter- and intra-observer agreement [[56], [57]] (with an intra-class correlation coefficient (ICC) of 0.98), its applicability is not as good as that of serum biomarkers. In the largest TE series reported to date (n = 13,369 examinations), failure to obtain any measurement has been reported in 3.1% of cases and unreliable results (not meeting manufacturer's recommendations) in 15.8% [58], mostly due to patient obesity or limited operator experience. Similar results have been reported in a large series of Asian patients (n = 3205) with failure and unreliable results rates of 2.7% and 11.6%, respectively [59].

An important question in clinical practice is whether unreliable results translate into decreased accuracy. It has been suggested that among the recommendations, the IQR/M <30% is the most important parameter for good diagnostic accuracy [[60], [61]]. In a recent study [62] in 1165 patients with chronic liver diseases (798 with chronic hepatitis C) taking liver biopsy as reference, TE reliability was related to two variables in multivariate analysis: the IQR/M and LS measure. Indeed, the presence of an IQR/M >30% and LS measure median ⩾7.1 kPa resulted in a lower accuracy (as determined by AUROC) than that of the whole study population and these cases were therefore considered “poorly reliable". Conversely, the highest accuracy was observed in the group with an IQR/M ⩽10% regardless of the LS measure. Also a recent study reported a significant discrepancy in up to 20% of cases in patients without cirrhosis between different FibroScan devices (402 vs. 502) [63]. These results require further validation before any recommendation can be made.

In order to minimize the number of patients with unreliable results due to obesity, a new probe (XL, 2.5 MHz transducer), allowing measurement of LS between 35 to 75 mm depth, has been developed [[64], [65], [66], [67], [68]]. Myers et al. [66]showed that in 276 patients with chronic liver disease (42% viral hepatitis, 46% NAFLD) and a BMI >28 kg/m2, measurement failures were significantly less frequent with the XL probe than with the M probe (1.1% vs. 16%; p <0.00005). However, unreliable results were still observed with the XL probe in 25% of case instead of 50% with the M probe (p <0.00005). Also it is important to note that stiffness values obtained with XL probe are lower than that obtained with the M probe (by a median of 1.4 kPa).

Apart from obese patients, TE results can also be difficult to obtain from patients with narrow intercostal space and are nearly impossible to obtain from patients with ascites [49]. As the liver is an organ with a distensible but non-elastic envelope (Glisson's capsule), additional space-occupying tissue abnormalities, such as edema, inflammation, extra-hepatic cholestasis, or congestion, can interfere with measurements of LS, independently of fibrosis. Indeed, the risk of overestimating LS values has been reported with other confounding factors including alanine aminotransferase (ALT) flares [[69], [70], [71]], extra-hepatic cholestasis [72], congestive heart failure [73], excessive alcohol intake [[74], [75], [76]], and food intake [[77], [78], [79], [80]], suggesting that TE should be performed in fasting patients (for at least 2 h) and results always interpreted being aware of these potential confounding [81]. The influence of steatosis is still a matter of debate with conflicting results: some studies suggest that steatosis is associated to an increase in LS [[82], [83], [84]] whereas others do not [[85], [86]].

Other liver elasticity-based imaging techniques

Several other liver elasticity-based imaging techniques are being developed, including ultrasound-based techniques and 3-D magnetic resonance (MR) elastography [87]. Ultrasound elastography can be currently performed by different techniques, which are based on two physical principles: strain displacement/imaging and shear wave imaging and quantification [88]. The latter allows a better estimation of liver tissue elasticity/stiffness, and includes point shear wave elastography (pSWE), also known as acoustic radiation force impulse imaging (ARFI) (Virtual touch tissue quantification™, Siemens; elastography point quantification, ElastPQ™, Philips) and 2D-shear wave elastography (2D-SWE) (Aixplorer™ Supersonic Imagine, France). pSWE/ARFI involves mechanical excitation of tissue using short-duration (∼262 μsec) acoustic pulses that propagate shear waves and generate localized, μ-scale displacements in tissue [89]. The shear wave velocity (expressed in m/sec) is measured in a smaller region than in TE (10 mm long and 6 mm wide), but the exact location where measurements are obtained can be selected by the operator under B-mode visualization. A major advantage of pSWE/ARFI is that it can be easily implemented on modified commercial ultrasound machines (Acuson 2000/3000 Virtual Touch™ Tissue Quantification, Siemens Healthcare, Erlangen, Germany; ElastPQ, iU22xMATRIX, Philips, Amsterdam, The Netherlands). Its failure rate is significantly lower than that of TE (2.9% vs. 6.4%, p <0.001), especially in patients with ascites or obesity [90]. Also its reproducibility is good, with ICC ranging from 0.84 to 0.87 [[91], [92], [93]]. However, like TE, pSWE/ARFI results are influenced by food intake [94] as well as necro-inflammatory activity and the serum levels of aminotransferases [95], both of which lead to an overestimation of liver fibrosis and have to be taken into account when interpreting the results. LS values obtained with pSWE/ARFI, in contrast to TE values, have a narrow range (0.5–4.4 m/sec). This limits the definitions of cut-off values for discriminating certain fibrosis stages and thus for making management decisions. Finally, quality criteria for correct interpretation of pSWE results remain to be defined.

2D-SWE is based on the combination of a radiation force induced in tissues by focused ultrasonic beams and a very high frame rate ultrasound imaging sequence capable of catching in real time the transient propagation of resulting shear waves [96]. The size of the region of interest can be chosen by the operator. 2D-SWE has also the advantage of being implemented on a commercially ultrasound machine (Aixplorer®, Supersonic Imagine, Aix en Provence, France) with results expressed either in m/sec or in kPa at a wide range of values (2–150 kPa). Its failure rate is significantly lower than that of TE [[97], [98], [99]], particularly in patients with ascites [[98], [99]], but not in obese patients when the XL probe is used for TE (10.4% vs. 2.6%, respectively) [100]. Similar to pSWE/ARFI, quality criteria for 2D-SWE remain to be defined.

MR elastography uses a modified phase-contrast method to image the propagation characteristics of the shear wave in the liver [101]. Elasticity is quantified by MR elastography (expressed in kPa) using a formula that determines the shear modulus, which is equivalent to one-third the Young's modulus used with TE [102]. The theoretical advantages of MR elastography include its ability to analyze almost the entire liver and its good applicability in patients with obesity or ascites. However, MR elastography remains currently too costly and time-consuming to be used in routine practice and cannot be performed in livers of patients with iron overload, because of signal-to-noise limitations.

Endpoints for staging liver fibrosis

In patients with viral hepatitis and HIV-HCV coinfection, the clinically relevant endpoints are: (1) detection of significant fibrosis (METAVIR, F ⩾2 or Ishak, ⩾3), which indicates that patients should receive antiviral treatment. However, with the availability of novel antiviral agents able to achieve sustained virological response (SVR) rates above 90% with limited side effects, it is likely that significant fibrosis will no longer represent an important decision making endpoint in HCV-infected patients. (2) Detection of cirrhosis (METAVIR, F4 or Ishak, 5–6) indicates that patients should not only potentially be treated for longer duration/different regimens in HCV but also monitored for complications related to portal hypertension (PH) and regularly screened for hepatocellular carcinoma (HCC). In NAFLD, representing another major etiology of chronic liver disease, the presence of significant fibrosis does not represent a relevant endpoint in the absence of standardized treatment regimens. However, detection of septal (advanced) fibrosis-cirrhosis seems clinically more relevant in NAFLD patients. In alcoholic liver disease (ALD), cholestatic liver diseases, and other etiologies, cirrhosis represents the most relevant clinical endpoint.

Performance of serum biomarkers for staging liver fibrosis

The diagnostic performances of serum biomarkers of fibrosis are summarized in Table 4. Overall, biomarkers are less accurate in detecting intermediate stages of fibrosis than cirrhosis. The most widely used and validated are the APRI (a free non-patented index) and the FibroTest® (a patented test that is not widely available), mainly in viral hepatitis C. A recent systematic review including 172 studies conducted in hepatitis C [103] reported median AUROCs of 0.79 and 0.86 for FibroTest® and of 0.77 and 0.84 for APRI, for significant fibrosis and cirrhosis, respectively. A meta-analysis by the developer [104] that analyzed data from 6378 subjects (individual data from 3282 subjects) who received the FibroTest® and biopsies (3501 with HCV infection and 1457 with HBV) found that the mean standardized AUROC for diagnosis of significant fibrosis was 0.84, without significant differences between patients with HCV (0.85) and HBV (0.80). Another meta-analysis [105] analyzed results from 6259 HCV patients from 33 studies; the mean AUROC values of APRI in diagnosis of significant fibrosis and cirrhosis were 0.77 and 0.83, respectively. Another meta-analysis of APRI in 1798 HBV patients found mean AUROC values of 0.79 and 0.75 for significant fibrosis and cirrhosis, respectively [106]. In the largest comparative study to date (n = 510 patients monoinfected with hepatitis B or C matched on fibrosis stage), overall diagnostic performances of blood tests (FibroTest®, FibroMeter®, and HepaScore®) were similar between hepatitis B and C with AUROC ranging from 0.75 to 0.84 for significant fibrosis, 0.82 to 0.85 for extensive fibrosis and 0.84 to 0.87 for cirrhosis, respectively [107].

Table 4
Diagnostic performance of serum biomarkers of fibrosis for significant fibrosis (F ⩾2) and cirrhosis (F4) in patients with chronic liver disease.

HCV, chronic hepatitis C; HBV, chronic hepatitis B; NAFLD, non-alcoholic fatty liver disease; AUROC, area under ROC curve; Se, sensitivity; Sp, specificity; CC, correctly classified: true positive and negative; n.a., not available.

∗∗HCV patients.

In HIV-HCV coinfected patients, performance of non-patented tests (e.g., APRI, FIB-4, and the Forns index) for predicting fibrosis seems less accurate than in HCV-monoinfected patients: they are accurate for the diagnosis of cirrhosis, but relatively inaccurate for the diagnosis of significant fibrosis [ [108], [109], [110]]. As for patented tests, such as FibroTest®, FibroMeter®, and HepaScore®, they outperform the non-patented tests in HIV-HCV coinfection, particularly for significant fibrosis [[111], [112]]. Importantly, one should be aware of false positive results with APRI and FIB-4 (related to HIV-induced thrombocytopenia) as well as with FibroTest® and HepaScore® (related to hyperbilirubinemia induced by the use of antiretroviral treatment such as atanazavir) or FibroTest® and Forns Index (related to increase in γ-glutamyl transferase induced by nevirapine) [111].

In patients with NAFLD, the NAFLD fibrosis score [40] is currently the most studied [[85], [113], [114], [115], [116], [117], [118]] and validated biomarker [119]. The NAFLD fibrosis score seems to perform better in Caucasians than Asians, probably related to the ethical difference in fat distribution and its influence on the BMI [102].

Comparative performance of patented and non-patented serum biomarkers for staging liver fibrosis

When compared and validated externally in patients with hepatitis C [[120], [121], [122], [123], [124], [125]], the different patented tests had similar levels of performance in diagnosis of significant fibrosis. In the largest independent study (1370 patients with viral hepatitis; 913 HCV and 284 HBV patients), which prospectively compared the widely used patented tests (FibroTest®, FibroMeter®, and HepaScore®) with the non-patented test (APRI), the AUROC values for significant fibrosis ranged from 0.72 to 0.78 with no significant differences among scores [124]. In patients with cirrhosis, the AUROC values were higher for all tests, ranging from 0.77 to 0.86, with no significant differences among the tests. Although non-patented tests such as the Forns index, FIB-4, and APRI were not as accurate as patented tests [125], there are no additional costs, they are easy to calculate, and are widely available.

Performance of TE for staging liver fibrosis

Performances of TE for diagnosing significant fibrosis and cirrhosis are summarized in Table 5 (viral hepatitis) & Table 6(non-viral hepatitis). The two index studies suggesting the interest of TE in the assessment of liver fibrosis have been conducted in patients with chronic hepatitis C [[126], [127]]. LS values strongly correlated with METAVIR fibrosis stages. However, it should be emphasized that despite high AUROC values, a substantial overlap of LS values was observed between adjacent stages of hepatic fibrosis, particularly for lower fibrosis stages. Many other groups have since confirmed these results [[86], [124], [125], [128], [129]], also in patients with hepatitis B [[69], [124], [129], [130], [131], [132], [133],[134], [135], [136], [137]] as well as in patients with HIV-HCV coinfection [[138], [139], [140], [141], [142], [143]].

Table 5
Diagnostic performance of TE for significant fibrosis (F ⩾2) and cirrhosis (F4) in patients with viral hepatitis B and C.

HCV, chronic hepatitis C; HBV, chronic hepatitis B; AUROC, area under ROC curve; Se, sensitivity; Sp, specificity; CC, correctly classified: true positive and negative; n.a, not available.

∗More than half of patients with «clinical» cirrhosis; adapted to ALT levels.

∗∗Validation cohort: HCV 92%; HBV 8%.

aAdapted to LT levels.

Table 6
Diagnostic performance of TE for F ⩾2 and F4 in chronic liver diseases other than viral hepatitis.

PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; NAFLD, non-alcoholic fatty liver disease; ALD, alcoholic liver disease.

AUROC, area under ROC curve; Se, sensitivity; Sp, specificity; CC, correctly classified: true positive and negative; n.a., not available.

TE is a reliable method for the diagnosis of cirrhosis in patients with chronic liver diseases, better at ruling out than ruling in cirrhosis (negative and positive predictive values 96% and 74%) [144]. TE more accurately detects cirrhosis (AUROC values, 0.80–0.99; correct classification ranging from 80% to 98%) than significant fibrosis (AUROC values, 0.65–0.97; correct classification from 57% to 90%) (Table 5 and Table 6). Several meta-analyzes [[145], [146], [147], [148], [149]] have confirmed the better diagnostic performance of TE for cirrhosis than for fibrosis, with mean AUROC values of 0.94 and 0.84, respectively [147]. In a recent meta-analysis of 18 studies including 2772 HBV patients [150], mean AUROC values for diagnosing cirrhosis and significant fibrosis were 0.93 and 0.86, respectively. However, we are still lacking a meta-analysis of data from individual patient data.

Different cut-offs have been proposed for cirrhosis according to etiologies ranging from 9.7 kPa in HBV [133] to 22.7 kPa in ALD [151]. However, it must be kept in mind that these cut-off values have been defined in a single population using ROC curves in order to maximize sensitivity and specificity – and not applied to a validation cohort. Difference between cut-offs may be simply related to difference in cirrhosis prevalence in the studied populations (ranging from 8% to 54%; Table 5, Table 6), known as the spectrum bias [[16], [17]]. Based on a meta-analysis, some authors have proposed an optimal cut-off of 13 kPa for the diagnosis of cirrhosis [147]. However, the cut-off choice must also consider the pre-test probability of cirrhosis in the target population (varying from less than 1% in the general population to 10% to 20% in tertiary referral centres). For instance, it has been shown that in a population with a pre-test probability of 13.8%, at a cut-off <7 kPa, cirrhosis probability ranged from 0% to 3% whereas at a cut-off >17 kPa cirrhosis probability was 72% [124].

When compared, the performances of TE have been shown to be similar between patients with HBV and HCV [[135], [136]]. Serum levels of aminotransferases should always be taken into account when interpreting results from TE, especially in patients with hepatitis B (who might have flares) [152]. To avoid the risk of false positive results, some authors have proposed to adapt TE cut-offs based on levels of ALT [132], a strategy that might not apply to patients with fluctuating levels of ALT or hepatitis flares (Table 5). Conversely, in hepatitis e antigen (HBeAg)-negative patients with normal levels of ALT, non-invasive methods, particularly TE, could be used as adjunct tools to measure HBV DNA, to follow inactive carriers or better identify patients who require liver biopsy (those with ongoing disease activity or significant fibrosis, despite normal levels of ALT) [[130], [153], [154], [155]].

TE has also been investigated in NAFLD patients but in a smaller number of studies [[66], [68], [82], [85], [156], [157], [158], [159]] (Table 6). Like in viral hepatitis, TE performances are better for cirrhosis than for significant fibrosis with AUROCs ranging from 0.94 to 0.99 and from 0.79 to 0.99, respectively. However, the performance of TE in NAFLD deserves several comments: Firstly, these studies have been conducted in heterogeneous and special populations such as Asian patients or children with low BMI (<28 kg/m2); secondly, most of them are underpowered with small sample size (<100 patients) and very few patients with cirrhosis; thirdly, the histological scoring systems such as those proposed by Brunt et al. [160] or Kleiner et al. [161] and endpoints (significant fibrosis or severe fibrosis) were heterogeneous in most studies evaluating fibrosis by TE in NAFLD. These differences in the study designs are likely the explanation for the observed differences among proposed cut-offs for a given endpoint (ranging for instance from 10.3 to 22.3 kPa for cirrhosis) (Table 6), known as the spectrum bias [[16], [17]]. Finally, all these studies have been conducted in tertiary referral centres with a higher proportion of patients with severe fibrosis than in the general population, making it difficult to extrapolate the performance of TE in detecting cirrhosis in large populations. Nevertheless, TE could be of interest to exclude confidently severe fibrosis and cirrhosis with high negative predictive value (around 90%) in NAFLD patients [85].

TE has also been evaluated in a variety of chronic liver diseases [[56], [144], [162]], as well as in primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC) [[163], [164]], and ALD [[151], [165]] (Table 6). However, in the latter it has been suggested by several groups that the presence of alcoholic hepatitis may influence LS results [[74], [75], [76]] and thus, TE should be ideally performed only after alcohol withdrawal in order to improve diagnostic accuracy.

Performance of other techniques for staging liver fibrosis

Point shear wave elastography using acoustic radiation force impulse quantification

Performances of pSWE/ARFI (Siemens) for diagnosing significant fibrosis and cirrhosis are summarized in Table 7. Most studies evaluated patients with mixed chronic liver disease with viral hepatitis being the predominant liver disease [[166],[167], [168], [169], [170], [171], [172], [173], [174], [175], [176], [177]]. Similar to TE, pSWE/ARFI more accurately detects cirrhosis (AUROC values: 0.81–0.99) than significant fibrosis (AUROC values: 0.77–0.94). The largest study evaluating pSWE/ARFI for staging of chronic hepatitis C was a retrospective pooled analysis of 914 international patient data [178], part of which were published in smaller single centre studies previously [[166], [167], [170], [171], [174], [179]]. It reported sensitivity and specificity of pSWE/ARFI for the diagnosis of significant fibrosis of 0.69 and 0.80 and for the diagnosis of liver cirrhosis of 0.84 and 0.76, respectively [178].

Table 7
Diagnostic performance of pSWE using ARFI for F ⩾2 and F4 in chronic liver diseases.

HCV, chronic hepatitis C; HBV, chronic hepatitis B; NAFLD, non-alcoholic fatty liver disease; AUROC, area under ROC curve; Se, sensitivity; Sp, specificity; CC, correctly classified: true positive and negative; n.a., not available.


∗∗Transformed in kPa.

Meta-analyzes have confirmed the better diagnostic performance of pSWE/ARFI for cirrhosis than for fibrosis [[180], [181]]. In a pooled meta-analysis including 518 individual patients with chronic liver disease (83% with viral hepatitis) mean AUROCs were 0.87 for the diagnosis of significant fibrosis, and 0.93 for the diagnosis of liver cirrhosis [180]. In a meta-analysis of 36 studies (21 full paper publications and 15 abstracts) comprising 3951 patients mean AUROCs were 0.84 (diagnostic odds ratio [DOR]: 11.54) for the diagnosis of significant fibrosis, and 0.91 (DOR: 45.35) for the diagnosis of liver cirrhosis [181]. Cut-off values suggested in the two meta-analyzes were 1.34–1.35 m/sec for the diagnosis of significant fibrosis and 1.80–1.87 m/sec for the diagnosis of cirrhosis. Only few studies have evaluated pSWE/ARFI in chronic hepatitis B [[182], [183]] and reported comparable results as for chronic hepatitis C and mixed chronic liver disease.

In a few studies pSWE/ARFI has also been investigated in NAFLD [[184], [185], [186], [187]]. Such as in viral hepatitis, pSWE/ARFI performances are better for severe fibrosis and cirrhosis than for significant fibrosis with AUROCs ranging from 0.91 to 0.98 and from 0.66 to 0.86, respectively. Interestingly, 80% of patients with BMI between 30 and 40 kg/m2 and 58% of patients with BMI >40 kg/m2 could be successfully evaluated using pSWE/ARFI [186]. Finally, pSWE/ARFI has also been evaluated in a variety of chronic liver diseases (ALD, PBC, PSC, and autoimmune hepatitis (AIH)). However, since most studies included mixed chronic liver diseases with predominantly viral hepatitis, the value of pSWE/ARFI for less common etiologies of chronic liver disease needs further evaluation.

2D-shear wave elastography

Only few studies [[96], [97], [188], [189]] have evaluated 2D-SWE for the staging of liver fibrosis, two of which used liver biopsy as reference method [[97], [189]]. In a pilot study in 121 patients with chronic hepatitis C (METAVIR fibrosis stage 41% F0/F1, 27% F2, 12% F3, and 20% F4), AUROCs of 2D-SWE for the diagnosis of significant fibrosis and cirrhosis were 0.92 and 0.98, respectively [189]. In another study in 226 patients with chronic hepatitis B (METAVIR fibrosis stage 17% F0, 23% F1, 25% F2, 20% F3, and 15% F4), 2D-SWE had AUROCS of 0.88 and 0.98 for the diagnosis of significant fibrosis and cirrhosis, respectively [97]. Sensitivities and specificities were 85% and 92% for the diagnosis of significant fibrosis using a cut-off of 7.1 kPa, and 97% and 93% for the diagnosis of cirrhosis using a cut-off of 10.1 kPa.

Other elastography methods such as strain elastography (a quasi-static technique) are available, but data for the staging of liver fibrosis are insufficient and seem to suggest that strain elastography has a worse diagnostic performance as compared to shear wave elastography [190].

Transient elastography vs. other techniques

Studies comparing TE and pSWE using ARFI show varying results. While many studies reported comparable results for both methods [[167], [174], [179], [191], [192]], some studies report better results for ARFI [172] and others better results for TE [[168], [174]], respectively. In a recent meta-analysis [90] including 13 studies (n = 1163 patients) comparing pSWE using ARFI with TE (11 full-length articles and two abstracts), no significant difference in DOR were found between ARFI and TE. Summary sensitivities and specificities for the diagnosis of significant fibrosis were 0.74 and 0.83 for ARFI and 0.78 and 0.84 for TE, respectively and 0.87 and 0.87 for ARFI and 0.89 and 0.87 for TE for the diagnosis of cirrhosis, respectively.

2D-SWE has been compared to TE in only three studies [[97], [100], [189]]. In chronic hepatitis C [189], AUROCs of SWE were significantly higher than with TE for the diagnosis of significant fibrosis (0.92 vs. 0.84, respectively; p = 0.002) but not for cirrhosis (0.98 vs. 0.96, p = 0.48). In chronic hepatitis B, AUROCs for SWE were significantly higher for both significant fibrosis (0.88 vs. 0.78) and cirrhosis (0.98 vs. 0.92) [97]. In 349 patients with chronic liver disease [100], SWE had a higher accuracy than TE for the diagnosis of severe fibrosis (⩾F3) (p = 0.0016), and a higher accuracy than pSWE using ARFI for the diagnosis of significant fibrosis (⩾F2) (p = 0.0003).

MR elastography has been compared to TE in patients with chronic liver diseases in three studies with conflicting results [[193], [194], [195]]. Two studies (a pilot Belgian study [193] and a Japanese retrospective study [195] in 96 and 113 patients with chronic liver disease) suggested that MR elastography might be more accurate than TE in diagnosis of significant fibrosis whereas another study from the Netherlands [194] in 85 patients with viral hepatitis reported similar accuracy for significant fibrosis. Further data are required to evaluate if MR elastography has superior accuracy for detecting significant fibrosis and cirrhosis as compared to TE, pSWE/ARFI, or 2D-SWE.

Comparison of performance of TE and serum biomarkers for staging liver fibrosis

Many studies have compared the performances of TE and serum biomarkers, mostly in viral hepatitis [[124], [125], [126],[143], [196], [197], [198], [199], [200], [201], [202], [203]] but also in NAFLD and ALD [[85], [165]]. TE and serum biomarkers have been shown to have equivalent performance for detecting significant fibrosis [[124], [125], [126]] but TE outperforms serum biomarkers for detecting cirrhosis [[124], [196], [199]]. However, given the lower applicability of TE (80% vs. 95% for serum biomarkers), performance could finally not differ for intention-to-diagnose analysis [125].

Algorithms combining different tests (LS and/or serum biomarkers)

Since the first proposal of a strategy combining TE and FibroTest® to increase diagnostic accuracy in patients with hepatitis C [126], many algorithms combining either TE and serum biomarkers [[125], [143], [198], [199], [200], [202], [204],[205]] or several serum biomarkers [[122], [206], [207], [208], [209], [210]] have been proposed, mainly in patients with viral hepatitis. Although these algorithms are more effective in detecting significant fibrosis than individual tests, they do not increase diagnostic accuracy for cirrhosis [[125], [196], [199]]. However, given the important clinical implications, in terms of prognosis, monitoring and treatment decisions that follow the diagnosis of cirrhosis, it seems justified to confirm a diagnosis of cirrhosis by two concordant but unrelated tests. Also ultrasound and other imaging methods hold a high specificity for the diagnosis of cirrhosis in this context, and may be useful as an unrelated method.

The advantage of combining two unrelated methods, such as TE and serum biomarkers, over the combination of two serum biomarkers is that TE provides more direct measurement of the liver structure than biomarkers, and that there is no relationship between the applicability of TE (success rate and interquartile range) and that of a biomarker [[204], [211]]. Also, the combination of TE and serum biomarkers might be more effective than the combination of two serum biomarkers for detecting significant fibrosis (significantly greater number of saved liver biopsies) [[200], [212]]. However, this strategy has only been validated in studies of patients with hepatitis C, is more costly, and could be hampered by the lower applicability of TE, compared with biomarkers. Most importantly, in case of unexplained discordance of non-invasive tests, a liver biopsy should still be performed.

Indications for non-invasive tests for staging liver disease in viral hepatitis

HCV including HIV-HCV

In the clinical management of HCV patients including those coinfected with HIV, there are several specific indications where the clinician can use non-invasive tests to aid in disease management. Either alone or in combination these tests allow for rapid staging of liver disease without the need for liver biopsy. The current gold standard for utilization of non-invasive tests to stage liver disease is to combine a serum biomarker with TE. The key for accuracy is to have concordance between the tests, which increases the diagnostic accuracy (Fig. 1). Every patient with chronic HCV infection should have liver disease staging at least once by non-invasive tests. Once a diagnosis of cirrhosis has been established, both AASLD and EASL guidelines recommend that those patients should be screened for PH and HCC [[213], [214]]. Therefore all HCV patients need to be staged as part of routine HCV care to exclude cirrhosis. The diagnostic accuracy of TE for cirrhosis has been confirmed by multiple studies and meta-analyzes and has proven superior to that reported by serum biomarkers.

Fig. 1
Proposed algorithm for the use of non-invasive tests in treatment-naive patients with Hepatitis C with or without HIV coinfection.

View Large Image | View Hi-Res Image | Download PowerPoint Slide

In chronic hepatitis B, TE generally has a higher AUROC as compared to serum biomarkers for advanced liver fibrosis [[198], [202]]. Among inactive carriers with normal transaminases, TE also has less fluctuation over time as compared to FibroTest® or APRI score [155]. LS of <5–6 kPa often indicates absent or minimal liver fibrosis [[132], [153]]. On the other hand, LS of >12–14 kPa often indicates liver cirrhosis (Table 5). Among patients with intermediate LS measurements, the accuracy of staging is lower. In doubtful cases, liver biopsy is recommended (Fig. 2). Among chronic hepatitis B patients who have elevated ALT levels or ALT flares, interpretation of LS measurement should be taken with caution. LS can be misleadingly high among patients who have severe acute exacerbation of chronic hepatitis B, even 3–6 months after ALT has been normalized [215].

Fig. 2
Proposed algorithm for the use of transient elastography in treatment-naive patients with Hepatitis B.

View Large Image | View Hi-Res Image | Download PowerPoint Slide

For HBeAg-positive patients, particularly among those who are older than 35 years of age with high normal ALT levels, non-invasive assessment of liver fibrosis is useful to differentiate whether patients are in immune tolerance phase or already have significant liver fibrosis secondary to immune clearance [216].

In HBeAg-negative patients, the low replicative phase is indicated by normal ALT level and low HBV DNA (<2000 IU/ml). On the other hand, the reactivation phase is characterized by elevated HBV DNA levels with intermittent elevation of ALT levels. Patients who have repeated and prolonged reactivation have higher risks of developing liver cirrhosis [217]. Non-invasive assessment of liver fibrosis is preferred over liver biopsy among HBeAg-negative patients with low (<2000 IU/ml) or borderline (>2000 to 20,000 IU/ml) HBV DNA and normal ALT levels, as the risks of advanced fibrosis and cirrhosis in these patients are usually below 10% [218].

Use of non-invasive tests for staging liver disease in NAFLD

NAFLD is a very common condition with reported prevalence of approximately 20% in different parts of the world [[219],[220]]. Simple steatosis does not increase mortality. Fibrosis is the most important prognostic factor in NAFLD and is correlated with liver-related outcomes and mortality [[2], [221]]. Advanced fibrosis, as determined by non-invasive serum biomarker, has been shown to predict liver-related complications and mortality [[222], [223]]. Not all NAFLD patients will develop advanced fibrosis. Biopsy series suggested a prevalence of advanced fibrosis in 50% of NAFLD patients [222], but a population-based study in Hong Kong revealed only 3.7% of the 264 NAFLD patients had advanced fibrosis [224]. NAFLD patients with metabolic syndrome and those with type 2 diabetes mellitus, had been shown to be at increased risk of having liver fibrosis in both Western and Asian cohorts [[220], [225]]. Fibrosis progression is possible among patients with simple steatosis or non-alcoholic steatohepatitis; approximately 25% to 37% of patients will have fibrosis progression in 3–5 years [[226], [227], [228]] [229]. Histologic inflammation and maybe metabolic factors are associated with higher risk of fibrosis progression among patients with simple steatosis or steatohepatitis [230].

Among the different serum biomarkers studied in NAFLD, only NFS and FIB-4 have been externally validated more than once, in different NAFLD populations and with consistent results [119]. These tests perform best at excluding severe fibrosis-cirrhosis (with negative predictive values >90%) and could therefore be used as a first line triage to identify patients at low risk of severe fibrosis. TE has excellent diagnostic accuracy for cirrhosis with a higher rate of false positive results than of false negative results and higher negative than positive predictive values. Therefore its ability to rule in severe fibrosis-cirrhosis may be insufficient for clinical decision making and may require histological confirmation.

Use of non-invasive tests for staging liver disease in other liver diseases

Alcoholic liver disease

Although the use of non-invasive tests in ALD has been explored, the methodological quality of existing studies is considerably heterogeneous without evaluation in large cohorts of ALD patients. Existing information on the usefulness of serum biomarkers has been recently summarized in the EASL guidelines for ALD and in recent reviews [[231], [232], [233]]. While a good performance has been reported for the use of FibroTest® in detecting significant fibrosis and cirrhosis (AUROC = 0.84 for F2-F4, AUROC = 0.95 for the diagnosis of cirrhosis), APRI has been found of limited use in the setting of ALD. Of note, FibroMeter® and HepaScore® have shown similar diagnostic accuracies than FibroTest® [234] with AUROC around 0.80 for significant fibrosis and 0.90 for cirrhosis. In addition, ELF® has also been shown to be useful in assessing fibrosis in ALD [34]. Interestingly, available data suggest that serum biomarkers of fibrosis may also be able to predict clinical outcomes [[234], [235]].

Information on elasticity-based techniques, mainly TE, in ALD is limited due to the scarcity of single-etiology studies. A recent systematic review from the Cochrane Collaboration, based on five retrospective and nine prospective cohort studies with a total of 834 patients, suggests that TE may be used as a diagnostic method to rule out severe fibrosis or cirrhosis in patients with ALD using cut-offs of 9.5 and 12.5 kPa, respectively [236]. However, the authors point out the risk of outcome reporting bias as well as caution on the use of currently recommended cut-offs as they are insufficiently validated and because there is the risk of overestimation of LS values in patients that are not abstinent from alcohol consumption.

Cholestatic liver disease

Available information regarding the use of non-invasive tests in cholestatic diseases is indeed more limited than that for viral hepatitis and NAFLD. This is due to the fact that patients with these diseases are usually part of cohorts of chronic liver disease and disease specific data on non-invasive tests performance is not reported separately. In the case of PBC, although histological proof of the disease is no longer considered mandatory to make the diagnosis, assessment of the disease stage remains useful for both prognostic (patients with more advanced disease have reduced survival than those in earlier stage) and therapeutic reasons (patients with earlier histological stage respond more favorably to UDCA administration and in patients with advanced disease surveillance of HCC is indicated) [237]. Thus, PBC patients often undergo a liver biopsy and the use of non-invasive tests of liver fibrosis may be advantageous in this setting. Several reports have tested the usefulness of serum biomarkers of liver fibrosis including serum levels of hyaluronic acid, procollagen III aminoterminal propeptide, collagen IV, and FibroTest® in patients with PBC [[238], [239], [240]]. While earlier studies did not provide estimates of diagnostic accuracy to readily assess clinical performance more recent reports do provide AUC values of ROC curves which in most of the cases are below 0.8. Thus, current evidence allows the conclusion that no single serum measurement has the ability to differentiate between early and advanced fibrosis in PBC [241]. In the case of PSC, no specific studies are available in this regard.

Reported data on the use of TE in PBC is encouraging. The report by Corpechot et al. of two cohorts of PBC patients evaluated with TE showed that this technique is currently one of the best surrogate markers of liver fibrosis in this disease [164]. This data is in agreement with findings from Floreani's group in Italy [242] and with an earlier study by Gomez-Dominguez from Spain [243]. In addition, TE may be useful to monitor PBC progression. In fact, prospective data from Corpechot et al. showed that progression of LS over time is predictive of poor outcome [164]. As for PSC, a recent study from the same group [244] showed that TE efficiently differentiates severe from non-severe liver fibrosis stages in this disease, and that both baseline LS measurements and increase over the time are able to predict patients' outcomes. Thus, TE seems to be a reliable non-invasive method for assessing biliary fibrosis in PSC patients [[163], [244]]. However, untreated dominant stricture of the common bile duct or primary hepatic ducts should be ruled out in PSC patients since obstructive cholestasis influences LS assessment [72]. Finally, with the availability of smaller probes (S1, S2), the use of TE has recently been tested in children with biliary atresia, a disease where fibrosis monitoring may help predict outcomes before surgery [245]. However, more data on non-invasive fibrosis evaluation in patients with cholestatic liver diseases is needed to make firm recommendations on the use of TE in this disease.

Autoimmune hepatitis

AIH may have insidious onset in a significant proportion of the cases, which result in a large number of cases (30% to 80%) being at the cirrhotic stage at the moment of diagnosis. Since a significant number of cases could be diagnosed without performing a liver biopsy [246], non-invasive tests for liver fibrosis may have a role in liver disease staging. Non-invasive tests could be also useful for monitoring response to immunosuppressive treatment in AIH, since fibrosis and cirrhosis can be reversible in this setting [[247], [248]]. However, specific data of either serum markers of fibrosis or imaging techniques is scarce to make recommendations. Of note, disproportionally high results of TE [249] have been reported in patients with AIH, which is likely related to the inflammatory activity considering that values decreased rapidly upon induction of disease remission.

Use of non-invasive methods when deciding for treatment in viral hepatitis

HCV including HIV-HCV

The current recommendations for treatment of HCV vary significantly between countries and healthcare systems according to the availability of therapy. However, the EASL and AASLD guidelines are clear as to the prioritization of treatment based on disease stage [[213], [214]]. There is some controversy in how best to use non-invasive tests in HCV therapeutic decisions. In countries where antiviral treatment is only indicated in patients with severe fibrosis or cirrhosis, both TE and serum biomarkers are effective – either alone, or in combination – to assess liver fibrosis. However the controversy is with significant fibrosis, where all staging parameters including non-invasive tests and liver biopsy have the greatest discordance and risk for inaccuracy. Since the diagnostic accuracy of non-invasive tests in differentiating between stage F1 and F2 has the most variability, this represents a challenge for clinicians [[17], [18]]. Although cut-offs for both TE and serum tests have been suggested for significant fibrosis, they have not been well validated and in a large, prospective US biopsy controlled study in over 700 HCV patients, TE, APRI, and FIB-4 all performed less well for significant fibrosis [129]. Combination of serum biomarkers with TE may marginally improve differentiation of F0–F1 from F2–F4 but has never been validated in actually differentiating between the single stages of F1 from F2.

In HIV-HCV coinfected patients, a priority has been given to treat all patients since this special population shows a more rapid disease progression so disease staging is less important for therapeutic decisions. However, in some countries anti-HCV treatment in HIV/HCV coinfected patients follows the same guidelines as for HCV monoinfection. There may be a reduced diagnostic accuracy of serum biomarkers for fibrosis in HIV-HCV patients and TE should be preferred.

There have been suggestions that as therapy becomes simpler and more effective with the advent of new direct-acting antiviral (DAA) agents and an increased uptake of HCV screening, that community based physicians, infectious disease physicians and internists may treat HCV [250]. In this situation, the role of non-invasive tests is also very important for determination of appropriate referral of patients with more advanced liver disease to specialists for appropriate disease monitoring [81].


Liver cirrhosis is the most important risk factor for liver-related complications and HCC in chronic hepatitis B. According to all regional guidelines, patients with liver cirrhosis and significant viraemia (HBV DNA >2000 IU/ml) should receive antiviral treatment regardless of the ALT levels [[251], [252], [253]]. Hence, non-invasive assessment of liver fibrosis can be considered in all patients in whom liver cirrhosis is suspected. Among hepatitis B patients who have elevated ALT but not yet reached two times ULN, liver fibrosis assessment can assist the decision of antiviral therapy. Patients who have significant liver fibrosis and HBV DNA >2000 IU/ml should be considered for antiviral therapy even if their ALT levels are below two times ULN [[251], [252]]. Among patients who have persistently elevated ALT >2 times ULN and HBV DNA >2000 IU/ml, all regional guidelines recommend commencement of antiviral therapy and liver fibrosis assessment may not be necessary.

Use of non-invasive methods for monitoring treatment response in viral hepatitis

HCV including HIV-HCV

A major advantage of non-invasive tests, compared with liver biopsy, is that they can be easily repeated over time in patients receiving antiviral therapy and that they could be used to monitor response to treatment and to evaluate fibrosis regression. Several studies reported a significant decrease in LS and biomarkers values, compared with baseline values, in patients with HCV who achieved SVR [[254], [255], [256], [257], [258], [259], [260], [261], [262], [263]], consistent with significant histologic improvement documented in studies of paired liver biopsies from HCV patients who achieved SVR [[264], [265]]. However, the changing levels of ALT and inflammation of successfully treated HCV patients can confound results of TE or biomarkers. Indeed, the major component of the significant decrease observed in LS and biomarkers values is not just reversal of fibrosis but also reduction in liver injury, edema and inflammation.

There are two important clinical questions about the use of non-invasive tests after antiviral treatment. First, what is the evidence of fibrosis and particularly cirrhosis reversal by non-invasive tests? The reversal of cirrhosis has important consequences in that it may alter long-term prognosis particularly for HCC occurrence in HCV patients and change the approach to screening and surveillance for HCC after SVR. This leads into the second question, what are the cut-off thresholds post SVR for determination of decreased risk of liver-related outcomes?

In HCV, there is only one study that has examined reversal of cirrhosis in 33 patients with cirrhosis with pre- and post-treatment liver biopsies and TE after SVR [266]. There was reversal of cirrhosis by biopsy in 19 patients with 11 of the 19 being METAVIR F3 and the remainder F1 or F2. Using a cut-off of 12 kPa, TE had a sensitivity of 61% and a specificity of 95%. The low sensitivity makes TE a poor tool to be utilized clinically as evidence of cirrhosis regression. Non-invasive tests, including TE and serum biomarkers, have been shown to predict liver-related outcomes in HCV patients [[267], [268]]. In both these studies clinical cut-offs of LS between 9.5 and 10.5 kPa were able to stratify patients at increased risk of clinical liver-related outcome. The best timing for repeated assessment of LS after therapy has not been established yet.


Prolonged treatment with antiviral therapy is associated with resolution of liver fibrosis and regression of liver cirrhosis [[269], [270], [271]]. Non-invasive tests are an attractive strategy to monitor changes in fibrosis. A significant decrease in LS and biomarkers values, compared with baseline values, in HBV-infected patients treated with analogs has been reported [[272], [273], [274], [275], [276], [277], [278], [279], [280], [281]]. However, like for HCV patients, improvement of LS in HBV patients who start antiviral therapy when ALT is elevated may be related to normalization of ALT instead of fibrosis improvement [274]. In this case, a LS measurement a few months after commencing treatment and normalization of ALT is recommended as baseline to monitor the changes in liver fibrosis. The value of non-invasive tests to predict the occurrence of complications or survival in patients with undetectable HBV DNA and cirrhosis prior to antiviral treatment remains to be determined [[282], [283], [284]].

Use of non-invasive tests for monitoring disease progression

Portal hypertension

There is substantial evidence indicating that TE can be quite effective in detecting patients with a high risk of having (or not having) developed clinically significant elevations of hepatic venous pressure gradient (HVPG) or varices. Several studies have shown that there is a good correlation between LS values and HVPG in patients with advanced liver diseases in both pre- and post-transplant settings [[285], [286], [287], [288]]. According to a recent meta-analysis, the diagnostic performance of TE for predicting clinically significant PH (CSPH, HVPG ⩾10 mmHg) in the setting of patients with compensated chronic liver disease/cirrhosis is excellent, with an AUROC of 0.93 [289]; a 90% sensitive cut-off for CSPH diagnosis is 13.6 kPa, and a 90% specific cut-off in this setting is 21 kPa. These cut-offs have been shown to allow a correct stratification of presence/absence of CSPH in patients with compensated cirrhosis and potentially resectable HCC, thus reducing the need for invasive hemodynamic assessment [290]. However, while the correlation is excellent for HVPG values between 5 and 10–12 mmHg (typical of cirrhosis without evident clinical manifestations related to PH), it hardly reaches statistical significance for values above 12 mmHg [286]. This is because, with the progression of cirrhosis, the mechanisms of PH become less and less dependent on the intrahepatic resistance to portal flow due to tissue fibrosis and progressively more dependent on extra-hepatic factors (i.e. hyperdynamic circulation, splanchnic vasodilatation) [291]. This observation sets a key limitation to the use of LS measurements as a non-invasive surrogate of HVPG beyond the prediction of clinically significant (HVPG ⩾10 mmHg) and severe (HVPG ⩾12 mmHg) PH, and, accordingly, TE of the liver is unlikely to be useful in monitoring the hemodynamic response to the administration of beta-blockers or disease progression in the decompensated phase. Conversely, repeated LS measurements could be useful during the first year after liver transplantation to identify patients with hepatitis C recurrence characterized by a rapid progression towards cirrhosis [292]; in addition, a LS ⩾8.7 kPa one year after orthotopic liver transplantation is associated to a worse prognosis in this setting [293].

More uncertain and controversial is the possibility of predicting the presence and the size of oesophageal varices (OV) based on LS values. A correlation between LS values and the presence of OV has been reported in several studies [[196],[286], [287], [288], [294], [295], [296]] with AUROCs ranging from 0.74 to 0.85 and cut-offs from 13.9 to 21.5 kPa. Although the sensitivity for the prediction of the presence of OV was high (76–95%), specificity was in general not satisfactory (43–78%). Regardless, the general features of these studies, i.e. single-centre retrospective, heterogeneous etiology of cirrhosis and stages of disease progression, subjective assessment of OV size, do not allow any sound conclusion on the utility of LS assessment in predicting the presence of OV and to screen cirrhotic patients without endoscopy [297].

Recently, studies employing different technical approaches have highlighted the potential utility of spleen stiffness (SS) assessment for predicting the presence of OV and the degree of PH in cirrhotic patients [[298], [299], [300], [301], [302]]. In particular, the study by Colecchia and co-workers [300] measured SS and LS by TE in 100 consecutive patients with HCV–induced cirrhosis. All patients also underwent measurements of HVPG and upper GI endoscopy. The ability of both SS and LS to predict clinically significant PH and the presence of OV was compared to that of the previously proposed methods, i.e. the LS–spleen diameter to platelet ratio score (LSPS) and platelet count to spleen diameter [[303], [304], [305]]. SS and LS were more accurate than other non-invasive parameters in identifying patients with OV and different degrees of PH. Further validation is needed before the place of SS in clinical practice can be defined.

Several biological parameters have been proposed for the non-invasive detection of CSPH including prothrombin time [287], a score combining platelet count and total bilirubin [306], and FibroTest® [307]. In particular, a score combining platelet count with total bilirubin had an AUROC of 0.91 for predicting CSPH with 88% sensitivity and 86% specificity at a cut-off of −1.0.

Similarly, several non-invasive tools have been proposed for the detection of OV including routine biological parameters [308], FibroTest® [309], and combination of simple biological and ultrasound parameters [303]. In the largest study to date comparing retrospectively a panel of serum markers (platelet count, AST/ALT ratio, APRI, Forns index, Lok index, FIB-4, and Fibroindex) in more than 500 patients with chronic liver diseases, the combination of Lok index (cut-off = 1.5) and Forns index (cut-off = 8.8) had the best diagnostic performance (AUROC of 0.80 and negative predictive value of 90%) for predicting clinically relevant OV [308].

In conclusion, the evidence accumulated so far indicates that both HVPG and upper GI endoscopy cannot be replaced by non-invasive methods, although an initial non-invasive approach may be helpful in selecting patients in whom these procedures are indicated with a certain level of urgency.

Hepatocellular carcinoma

To date, several cross-sectional studies [[310], [311], [312], [313]] identified that high LS value measured by TE is significantly associated with the risk of presence of HCC (Table 8). However, these cross-sectional studies only describe the 'static' phenomenon that patients with HCC have high LS values than those without HCC, not considering the 'dynamic' association between the progression or regression of liver fibrosis and the risk of future HCC development. To overcome this limitation, several longitudinal prospective studies [[267], [314], [315], [316], [317], [318], [319], [320], [321],[322], [323]] have recently been published (Table 8).

Table 8
Prognostic performance of TE for predicting development of HCC in patients with chronic liver disease.

^Liver cirrhosis with Child-Pugh class A. ∗Mostly HCV with 85.4%. ∗∗41 patients with HCC at the time of enrollment, 10 patients developed HCC during follow-up.

†All patients treated with interferon; ‡All patients were HBeAg-negative; ¥All patients completed 2-year entecavir treatment.

TE, transient elastography; HCC, hepatocellular carcinoma; AUROC, area under the receiver operating characteristic curve; kPa, kilopascal; HCV, hepatitis C virus; HBV, hepatitis B virus; n.a., not available.

A large prospective cohort study with chronic hepatitis C (866 patients) was conducted in Japan [317]. Along with age, male gender, and clinical cirrhosis, stratified LS values were identified as an independent risk factor for HCC development. Compared with patients with LS values ⩽10 kPa, patients with higher LS values were at significantly increased risk of developing HCC (LS values, 10.1–15 kPa, hazard ratio [HR], 16.7; LS values, 15.1–20 kPa, HR, 20.9; LS values, 20.1–25 kPa, HR, 25.6; and LS values, >25 kPa, HR, 45.5). In addition, the cumulative incidence rates of HCC showed a stepwise increase according to the stratified LS values (p <0.001 by the log-rank test). In addition, Jung et al. [316] further validated the usefulness of TE in prediction of HCC development in patients with chronic hepatitis B (n = 1130). Compared with patients with LS values ⩽8 kPa, patients with higher LS values were at significantly increased risk of developing HCC (LS values, 8.1–13 kPa, HR, 3.07; LS values, 13.1–18 kPa, HR, 4.68; LS values, 18.1–23 kPa, HR, 5.55; and LS values, >23 kPa, HR, 6.60). Furthermore, changes in the risk of HCC development according to the pattern of changes in the measured LS was also shown in the study, suggesting a potential role for serial LS measurement as a dynamic monitoring tool for risk estimation of HCC development in patients with chronic hepatitis B. All these results implied that TE is useful in estimating the risk of HCC development in patients with chronic liver disease across the etiology despite different carcinogenetic mechanisms of HCV and HBV.

Based on the close relationship between TE and the risk of HCC development, several studies have tried to develop and validate LS-based prediction models for HCC development [[310], [320], [324]]. Wong et al. [324] evaluated the accuracy of LS-HCC score, refined from their previous CU-HCC score [325] in 1555 Asian patients with chronic hepatitis B. The AUROC of LS-HCC score was higher than that of CU-HCC score in predicting HCC development (0.83 vs. 0.75 at 3 year, 0.89 vs. 0.81 at 5 year). More recently, Kim et al. [320] also introduced a predictive model based on a Cox proportional hazards model using age, male gender, LS value, and HBV DNA in patients with chronic hepatitis B. This model showed good discrimination capability, with an AUROC of 0.806 (95% CI 0.738–0.874) and AUROC remained largely unchanged between iterations, with an average value of 0.802 (95% CI 0.791–0.812). The predicted risk of HCC occurrence calibrated well with the observed risk, with a correlation coefficient of 0.905 (p <0.001).

According to all the results regarding non-invasive markers, it can be concluded that non-invasive methods are not merely an alternative to biopsy for staging fibrosis, but also predictive of the incidence of liver-related complications of liver fibrosis, including HCC development. However, further studies focusing on diverse etiologies of chronic liver disease, such as ALD or NAFLD, are needed to expand the clinical prognostic usefulness of non-invasive methods. In addition, optimal cut-off values with respect to the different etiologies of chronic liver disease to assess the risk of HCC development should be set up in subsequent larger longitudinal prospective studies. In spite of several limitations, non-invasive methods to assess and monitor the risk of HCC development will help physicians to establish optimum treatment strategies. It should be further investigated whether the accuracy of the surveillance strategy can be enhanced, if these non-invasive methods are incorporated into the routine surveillance strategy.

Determining prognosis

There is increasing evidence for the prognostic value of non-invasive tests in patients with chronic liver diseases. Several recent studies have shown that in patients with chronic liver disease, LS could also predict clinical decompensation as well as survival [[244], [268], [282], [321], [322], [326], [327], [328]]. For instance, Robic et al. [322] found that TE was as effective as HVPG in predicting clinical decompensations in 100 patients with chronic liver disease with a 2 year follow-up. Both HVPG ⩾10 mmHg and LS ⩾21.1 kPa had 100% NPV for portal-hypertensive complications. Similarly, in a cohort of 128 Korean patients with active HBV cirrhosis, LS at a cut-off of 19 kPa, had a hazard ratio of 7 for development of clinical decompensation [321]. In a cohort of 1457 HCV patients, LS values and FibroTest® had the highest 5 year predictive values for predicting survival and liver-related death, which did not change after adjustment for treatment response, patient age, and estimates of necro-inflammatory grade [268]. Interestingly, Corpechot et al. [244] have shown in 168 patients with PSC that, not only those with high baseline but also those with increase in LS values (>1.5 kPa/year) were at a very high risk (approximately 10 times the risk estimated in the other group) of death, liver transplantation, or hepatic complications within a 4 year period. In another study in 1025 patients with chronic hepatitis C, the prognosis of patients with LS between 7 and 14 kPa on inclusion was significantly impaired when an increase ⩾1 kPa/year was observed [263].

Finally, it has been recently suggested that SS could predict the occurrence of complications [329]. Thus the potential of LS values for predicting clinical outcomes seems to be greater than that of liver biopsy, probably LS measures ongoing pathophysiological processes and functions that a biopsy cannot.

Similarly, serum biomarkers such as FibroTest® [[154], [234], [330]], ELF® [[235], [239]], APRI and FIB-4 [[222], [331]], as well as for models based on standard laboratory tests [[332], [333]] have been shown to have prognostic value in various chronic liver diseases.