Transient elastography (TE) (Fibroscan^®-Echosens, Paris, France) is a non-invasive technique developed in 2003. It was conceived to indirectly assess liver fibrosis by measuring liver stiffness (LS)[6]. It is performed with an ultra-sound transducer probe mounted on the axis of a vibrator that produces vibrations of mild amplitude and low frequency. This induces an elastic shear wave that propagates through the underlying liver tissues. The velocity of the shear wave is directly related to tissue stiffness: the harder the tissue, the faster the shear wave propagates. TE explores a volume of liver parenchyma which is approximately 1/500 of the total liver mass, at least 100 times bigger and far more representative than a biopsy sample. The examination is painless, rapid, user-friendly and easy to perform at the bedside or in the outpatient clinic[7]. It is worth stressing though that the clinical interpretation of TE results should always be carried out by an expert clinician, aware of patient demographics, disease aetiology and routine laboratory parameters[7]. In fact, an Italian core group study has recently highlighted the need for a very careful quality control in order to better apply TE to clinical practice, with special attention to a patient’s conditions and to the success rate of the stiffness measures. That study has also recommended the use of a detailed report form for routine TE results[8].

TE has been demonstrated to be a highly reproducible technique in terms of inter- and intra-observer agreement with intra-class correlation coefficients of 0.98[9].

Absolute contra-indications to TE examination are pregnancy and the presence of implantable devices, such as pacemakers or defibrillators. Liver stiffness measurement (LSM) can be difficult or even impossible in obese patients or in those people with narrow intercostal space or ascites[6]. Several studies in this field have reported failure rates ranging from 2.4% to 9.4%[6,9-16]. Foucher et al[16] reported that the only factor associated to failure was a body mass index above 28 (odds ratio 10.0; 95%CI: 5.7-17.9, P = 0.001). Further experience has allowed to indicate that, rather than the body mass index, a limiting factor for the success rate may be a fatty thoracic belt[7]. In fact, in over-weight or obese patients the fatty thoracic belt seems responsible for the attenuation of both elastic waves and ultra-sound, making LSM impossible. Dedicated probes and new algorithms based on the attenuation of both ultra-sonic and shear waves have been proposed with some benefits on TE accuracy[10,17].

Some studies examined LS values in apparently healthy subjects[18,19] leading to comparable results. In fact, the mean LS values reported by these studies were superimposable and ranged from 4.8 to 6.9 kilopascal (kPa). In addition, LS values were not influenced by age but higher values were reported in men as compared to women and in the presence of steatosis or features of the metabolic syndrome[18,20].

Inactive HBV carriers are defined as hepatitis B e antigen (HBeAg) negative chronic HBV carriers with normal alanine aminotransferase (ALT) and HBV-DNA persistently at < 2000 international units/mL. Some studies have assessed LS values in inactive HBV carriers showing results superimposable to those of healthy subjects. Oliveri et al[21] measured the LS value in 68 inactive HBV carriers. The mean stiffness value was 5.0 ± 1.8 kPa, with a significant difference between subjects with abnormal ALT and steatohepatitis or steatosis at histology (n = 17, 6.9 ± 2.3 kPa), and subjects with normal ALT and without dysmetabolic profile (n = 57, 4.3 ± 1.0 kPa). Similar results have also been obtained by Maimone et al[22] who focused on the usefulness of TE in the discrimination between HBeAg-negative disease and inactive HBeAg-negative carriers. TE was performed in 220 subjects, of whom 125 inactive carriers. LS values resulted significantly lower in the inactive carriers group than in the chronic disease group, with the inactive carriers showing a mean LS value of 4.8 ± 1.2 kPa and a median value of 4.7 kPa (range: 2.4-7.9 kPa).

Castéra et al[23] evaluated 201 inactive HBV carriers by means of TE, obtaining LS values (median 4.8 kPa vs 6.8 kPa, P < 0.0001) significantly lower than in 128 CHB patients, without differences related to HBV DNA levels. Among them, 82 inactive carriers could be evaluated with repeated TE within a median of 21.7 mo (range: 3.3-49.1 mo), showing no significant variation in LS values over time [median intra-patient changes at the end of follow-up -0.2 kPa, range: (-1.2)-(+0.7), P = 0.12], pointing to TE as a reliable tool for the follow-up of inactive HBV carriers and for a better selection of patients requiring liver biopsy.

In a recent study[24] performed on a cohort of 361 HBeAg negative HBV carriers who underwent TE at baseline and 3 years later, liver fibrosis progression - which was arbitrarily defined as “an increase in LS by at least 30% to a value suggestive of advanced fibrosis” according to internal diagnostic algorithms - appeared to be rare among patients with a HBV DNA level < 20000 UI/mL (2.8%) and extremely rare in inactive HBV carriers (0.8%)[24].

In CHB the assessment of the extent of fibrosis is crucially important to take appropriate therapeutic decisions on whether to start antiviral treatment or to evaluate the need for HCC surveillance. Because of the complex natural history of CHB, which frequently displays fluctuating patterns in term of necroinflammation, differences evidently exist between CHB and other chronic liver diseases such as chronic hepatitis C (CHC). Several studies have analysed the diagnostic accuracy of TE in predicting the stage of fibrosis in CHB patients: the main results are summarized in Table 1. For the purpose of diagnosing significant fibrosis (i.e., F ≥ 2 or S ≥ 3) the LS cut-off used ranged from 5.2 to 8.7 kPa. The sensitivity estimates ranged from 70% to 93% and specificity estimates from 38% to 92%. Corresponding positive and negative likelihood ratios (LR⁺ and LR^-) ranged from 1.4 to 8.2 and from 0.07 to 0.40, respectively. For the purpose of diagnosing liver cirrhosis, the proposed LS cut-off values varied from 10.3 to 13.4 kPa with sensitivity estimates ranging from 59% to 100% and specificity from 79% to 94% and corresponding LR⁺ and LR^- from 0.07 to 0.3 and from 2.0 to 9.9 respectively, as shown in Table 2.

These wide variations in diagnostic accuracy estimates among the studies available in the literature can be related to differences in the spectrum of the population analysed and in the prevalence of the disease and particularly to the different LS cut-off values used. In fact, the main limitation to the use of the technique in clinical practice is the definition of LS cut-off values that help reliably classify patients in a given fibrosis stage: it is widely known that a substantial degree of overlap exists among patient groups of adjacent fibrosis stages. Usually, the cut-off value is derived from the receiver operating characteristics (ROC) curve and it is the one optimizing both sensitivity and specificity. An available different approach is about choosing two different cut-offs, one that maximizes sensitivity to exclude the diagnosis and the other that maximizes specificity for a confirmation strategy. Viganò et al[25] validated an algorithm with two distinct cut-offs for the positive and negative prediction of significant fibrosis and cirrhosis in treatment-naïve patients with CHB. Patients were examined by percutaneous liver biopsy and TE: a > 13.1 kPa positive and a ≤ 9.4 kPa negative cut-off for cirrhosis had a > 90% sensitivity and specificity, with an overall accuracy prediction of 94%, independent of alanine aminotransferase values.

Only a few studies have carried out direct comparisons between CHB and CHC patients. Both Cardoso et al[26] and Verveer et al[27] have assessed - in a cross-sectional study - treatment-naive patients with CHB or CHC who underwent TE measurement and liver biopsy, showing that the overall TE diagnostic performance was similar in the two patient groups. In the meta-analysis by Tsochatzis et al[28] the reported LS cut-offs were globally lower in CHB as compared to CHC group (on average 7.0 vs 7.6 for F ≥ 2, 8.2 vs 10.9 for F ≥ 3 and 11.3 vs 15.3 for F4). The authors have suggested that a possible logical explanation for this finding is that fibrosis septa are usually thinner in CHB patients than in those with CHC. In addition, because CHB tends to result in cirrhosis with larger nodules, i.e., macro-nodular cirrhosis, than CHC does, TE waves are more likely to pass the normal liver parenchyma between fibrotic bands leading to lower stiffness values.

Liver stiffness is a physical parameter primarily related to fibrosis, but it can also be influenced by other factors that modify liver elasticity, such as inflammatory infiltrate variations[15], oedema, vascular congestion[29], cholestasis[30] and, even if still controversial, hepatic steatosis[31]. Some earlier reports have suggested the influence of steatosis on LS be negligible but this association has possibly been attenuated by the low prevalence of steatosis[32]. Data from literature seems to show a poorer association between steatosis and CHB than between steatosis and CHC, with lower LS values and corresponding cut-offs in CHB. By contrast, liver inflammation clearly contributes to LS and may be more pronounced in CHB patients, especially during a hepatitis flare or viral reactivation. In their prospective study Verveer et al[27] confirmed that necroinflammation may induce higher LS values in both CHC and CHB. A demonstration that the presence of elevated ALT levels can increase LS values descends from the evidence that after the introduction of an antiviral therapy there is a rapid decrease of LS in parallel with ALT normalization[33]. A question remains unsolved about whether ALT levels, which frequently fluctuate in CHB, should be taken into consideration to adapt the cut-offs used for LS categorization: Chan et al[34] have proposed an algorithm which uses higher cut-offs in patients with elevated ALT levels, to avoid the overestimation of fibrosis because of the inflammation. Interestingly, adapting cut-offs on ALT did not improve the overall percentage of patients correctly classified. From all these data one can possibly conclude that, since LS is the consequence of multiple intra-hepatic events, including steatosis and inflammation, it will never be a perfect marker of fibrosis by itself, especially in the context of CHB but it can more usefully used as an overall marker of hepatic “well-being”.

In a study by our own group[31], we have made a comprehensive evaluation of the factors that would account for any discrepancies in diagnostic accuracy between TE and the standard-of-care liver biopsy in CHB as compared to CHC patients. The results of the study evidenced that fibrosis stage and liver cell necroinflammatory activity were independently associated with TE results in both HBV and HCV patients, whereas steatosis was independently associated with TE only in HCV. Fibrosis over-estimation was predicted by severe/moderate necroinflammatory activity in HBV and by older age (41-60 years or > 60 years vs < 40 years), > 2 upper limit of normal (UNL) AST and > 2 UNL GGT, as well as severe/moderate necroinflammatory activity and severe/moderate steatosis in HCV.

Patients with cirrhosis need strict follow-up for the surveillance of hepatic decompensation, portal hypertension and its complications and the early detection of HCC. Hepatic vein pressure gradient (HVPG) and upper gastro-intestinal endoscopy are the current reference standards in the assessment of portal hypertension, both techniques being expensive and invasive. Several groups have tried to develop non-invasive methods to stage the grade of portal hypertension and possibly predict its complications. Literature data suggest TE be a simple and reliable method to use in every-day clinical practice[35].

TE has showed a good correlation with HVPG in different studies[36-40]. The main objectives on which the studies focused have been the detection of clinically significant portal hypertension and the presence of esophageal varices (EV): in CHC and other aetiology settings, TE accuracy seems to be satisfactory in predicting significant portal hypertention whereas its diagnostic performance in predicting the presence and severity of EV is lower and often sub-optimal.

In HBV settings some groups, mainly from Asia, have tried to assess the role of liver stiffness measurement by TE in predicting the presence of varices in cirrhotic patients, especially high-grade EV. Data are summarized in Table 3. Chen et al[41] performed both TE and upper endoscopy in 238 HBV cirrhotic patients. As TE alone could not predict EV with satisfactory efficiency (area under receiver operating characteristic curve, AUROC 0.73, 95%CI: 0.66-0.80), they tried to create a novel algorithm to improve the performance: USLS, an ultrasonic score calculated on the morphologic parameters reflecting hepatic fibrosis progression. The AUROC comparison thus showed an improved performance in predicting the presence of high-grade EV in the sub-group of patients with ALT > 5 ULN (0.85, 95%CI: 0.76-0.94). Kim et al[42] enrolled and evaluated 401 consecutive CHB patients, developing a prediction model to detect high-grade EV by means of combining LS, spleen diameter and platelet count, showing a good overall accuracy (AUROC 0.953). The same group[43] assessed the cumulative risk of future EV bleeding in 577 consecutive HBV cirrhosis patients without history of EV bleeding. The combination of LS, spleen diameter and platelet count was useful in identifying a sub-group of patients at significantly higher risk of bleeding among the patients with so-defined high-risk EV (P < 0.001).

LS has been evaluated also as a possible predictor of the development of HCC. Several groups assessed this aspect, mainly in CHC settings, showing a good correlation between high basal values of LS and the increased risk of HCC development during follow-up[44-46].

In HBV settings Chon et al[47] compared the accuracy of several non-invasive liver fibrosis panels [age-spleen-platelet ratio index, LS, SSM; LS-spleen diameter-platelet ratio index (LSPI), P2/MS, and fibrosis score 4 (FIB-4)] in predicting the development of HCC or hepatic decompensation - defined as variceal bleeding, ascites, hepatic encephalopathy, spontaneous bacterial peritonitis, or hepatorenal syndrome - in a large cohort of CHB patients (n = 1126). During a median follow-up of 30.7 mo the accuracy of LS values and LSPI in predicting the development of HCC or hepatic decompensation was higher than that of other parameters, although with a relatively low predictive capacity (AUROC = 0.789 and 0.788 vs 0.729, 0.756, 0.696, 0.744 for HCC development; AUROC = 0.820 and 0.848 vs 0.787, 0.799, 0.812, 0.784 for hepatic decompensation). By multivariate analysis including demographic and laboratory data, LS was found to be an independent predictor of HCC development (HR 1.040; 95%CI: 0.012-1.070, P = 0.006) together with age, male gender, and HBeAg positivity. Both LS (HR 1.033 95%CI: 1.007-1.060, P = 0.013) and LSPI (HR 1.002; 95%CI: 1.001-1.004, P = 0.003) were independent predictors of hepatic decompensation.

In a separate study[48] 128 patients with CHB underwent TE and liver biopsy before starting nucleot(s)ide analogues (38.3% lamivudine, 61.7% entecavir); all the patients were viremics (HBV DNA > 2000 IU/mL) with histological fibrosis ≥ F3. All the patients were followed up regularly to detect decompensation (variceal bleeding, ascites, hepatic encephalopathy, spontaneous bacterial peritonitis, hepatorenal syndrome) and HCC. During a median follow-up period of 27.8 mo, 19 patients developed major events (14.8% of all patients): hepatic decompensation (n = 5), HCC (n = 13) and both conditions (n = 1). Serum albumin, prothrombin time, and platelet count were significantly higher in those patients without liver related events; AFP, and LSM values were significantly higher among those who developed liver-related events. Multivariate analysis including demographic data, laboratory tests, liver biopsy data, type of antiviral treatment and HBV DNA levels after treatment initiation, identified LS as the only independent predictor of liver related events (HR 1.038; 95%CI: 1.002-1.081, P = 0.044), especially for patients with LS ≥ 19 kPa (HR 7.176; 95%CI: 2.257-22.812, P = 0.001).

Finally, also the large study by Jung et al[49] assessed the new role of TE in predicting the risk of HCC development in 1130 patients with CHB. Between May 2005 and December 2007, 672 (59.5%) patients were treated with antiviral drugs before or after enrolment. All 1130 patients underwent LSM with a median value of 7.7 kPa and then attended regular follow-up for HCC surveillance. During a median period of 30.7 mo (range: 24.0-50.9 mo), 57 patients developed HCC with cumulative incidence rates of HCC development in the first, second and third year of 0.80%, 3.26%, and 5.98%, respectively. The data collected showed that the patients with higher LS value (> 8 kPa) were at a significantly greater risk of HCC development: HR were: 3.07 (95%CI: 1.01-9.31, P = 0.047) for LS 8.1-13 kPa, 4.68 (95%CI: 1.40-15.64, P = 0.012) for LS 13.1-18 kPa, 5.55 (95%CI: 1.53-20.04, P = 0.009) for LS 18.1-23 kPa and 6.60 (95%CI: 1.83-23.84, P = 0.004) for LS > 23 kPa, encouraging the regular use of TE in clinical practice to stratify the risk of development of HCC also in the HBV subset.

Some studies in HCV settings[38,50] have recently showed that TE as well as other non-invasive tests can predict clinically relevant outcomes, including survival, even better than liver biopsy[50].

In HBV setting de Lédinghen et al[51] first investigated the prognostic role for liver fibrosis of TE and other non-invasive tests (FibroTest, APRI, FIB-4) as compared to liver biopsy. The Authors assessed the 5-year overall survival in a consecutive cohort of 600 patients with CHB (inactive carriers 36%, n = 209). After a 5 years long observation period they had 25 deaths of which 13 for liver-related factors and four patients underwent OLT. Among inactive carriers no liver-related deaths were observed. Survival without liver-related death was 96.3%, overall survival 94.1%, and it resulted significantly low in patients with severe fibrosis, detected by both non-invasive method (P < 0.0001) and liver biopsy (P = 0.02). At year 5, 97.1% of patients with a LSM < 9 kPa was alive vs 38.5% of patients with a LSM > 20 kPa. In the multivariate analysis, FibroTest and LS were both independent predictors of survival: LS showed a HR of 6.8 (1.6-28.7), P < 0.001 with a prognostic AUROC of 0.80 (0.70-0.87).

HBV-related liver cirrhosis is a complex clinical field, in which recent clinical trials have suggested that CHB treatment may lead to cirrhosis improvement. A recent study was performed on CHB patients with advanced fibrosis or cirrhosis prior to treatment with nucleoside/nucleotide analogue for at least one year; they underwent LS measurement by means of TE at follow-up[52]. LS cut-off levels for diagnosing fibrosis stage F2, F3 and F4 were ≥ 7.2 kPa, ≥ 8.1, and ≥ 11.0 kPa, respectively. Eighty percent of the patients had cirrhosis and 20% advanced fibrosis (F3) prior to treatment. The median treatment duration was 50.5 mo. Among the patients with cirrhosis prior to treatment, 26 (49%) had LS below 11.0 kPa at follow-up, suggesting cirrhosis regression. Among the patients with advanced fibrosis (F3) prior to treatment, 10 (77%) had LS below 8.1 kPa after treatment, suggesting fibrosis improvement.

Acoustic radiation force impulse: Acoustic radiation force impulse (ARFI) quantification (Virtual Touch™ Siemens ACUSON S2000™, Siemens Medical Solution, Mountain View, CA, United States; ElastoQ Philips iU22, Philips, Bothell, United States) is obtained by means of short-duration acoustic pulses around 2.6 mHz. The pulses induce the compression of the tissue and consequently the development of shear waves that propagate at a certain velocity proportional to tissue stiffness; the propagation velocity of the shear wave is detected by the ultrasound probe and is displayed on the screen. Basically, the speed of the shear wave propagation correlates with the stiffness of the tissue and thus with the severity of liver fibrosis in the context of liver disease[53]. The advantage of this technique is that ARFI can be included in the standard B-mode liver ultra-sound equipment, becoming an easy immediate complement to abdominal ultra-sound examination.

Data on several patients with chronic liver disease of different aetiology show that ARFI quantification has a good performance for liver fibrosis staging with slightly better results with regard to the assessment of severe fibrosis (AUROC 0.91) and cirrhosis (AUROC 0.93) than for significant fibrosis, coherently with the results of TE. Proposed optimal cut-offs are 1.34 m/s for the diagnosis of significant fibrosis, 1.55 m/s for severe fibrosis and 1.8 m/s for cirrhosis[54]. A recent study assessed ARFI in 114 CHB patients undergoing liver biopsy and showed a significant correlation between ARFI quantification and histological fibrosis stage. The overall diagnostic accuracy expressed as areas under ROC curves for ARFI imaging was 0.75 for the diagnosis of significant fibrosis (F ≥ 2), 0.93 for severe fibrosis (F ≥ 3) and 0.97 for cirrhosis, without any significant difference between ARFI and TE performed in 92 out of 114 patients, thus confirming the reliability of this technique also in the HBV setting[55]. To our knowledge, there is still a lack of data to support the prognostic value of ARFI quantification of LS in terms of prediction of mortality or liver-related complications, especially in chronic HBV settings.

Magnetic resonance elastography (MRE) is a MRI-based technique which has been extensively evaluated and demonstrated as an accurate predictor of liver fibrosis with very high diagnostic performances[56-60].

Most of the studies were performed in populations with chronic liver disease of various aetiologies, such as CHC and alcohol-related cirrhosis and NASH. One study by Venkatesh et al[61] assessed the performance of MRE in 63 CHB patients, showing a good performance in the diagnosis of significant fibrosis (AUROC 0.99), and cirrhosis (AUROC 0.98).

Very recently a further promising use of TE has been found to be the measurement of spleen stiffness (SS) in assessing liver disease severity in CLD patients, considering the well-known role of splenomegaly as an indicator of cirrhosis, related to portal hypertension and splenic congestion in increased vascular resistance, leading to fibrosis of the spleen. In the last four years many studies have focused on this application of TE or ARFI elastography in clinical practice[62-64]. In CHC settings the correlation of SS with fibrosis stage has been assessed, showing a good correlation between SS values > 60 kPa and the presence of cirrhosis[63]. PH and varices presence have also been evaluated in this setting, showing a significant correlation between HVPG values and both LS and SS and an association between SS values and EV presence. Data from our group[62] suggest a good inter-observer agreement for SS measurement in chronic hepatitis settings and a positive correlation between SS, as well as LS, and EV presence, thus suggesting that the combined use of the two parameters can reliably predict the presence of varices and be useful in assessing liver disease severity.

Most studies on SS assessment have been conducted in HCV settings, but the interest in its role for HBV positive patients is increasing. The main studies conducted in this population have measured SS using ARFI technology.

In fact, preliminary results indicated that the ARFI measurement of SS can be used as a non-invasive tool to rule out cirrhosis (AUROC = 0.91, accuracy = 87.1%) and to predict the presence of EV[65-67].

Chen et al[68] evaluated the diagnostic role of SS measurement, using ARFI technology, for liver fibrosis assessment in patients with CHB (n = 61) or C (n = 102), performing also LSM and percutaneous liver biopsy. Intra-observer correlations coefficients (ICC) were evaluated: ICC for LSM and SSM were 0.993 (95%CI: 0.981-0.997, P < 0.001) and 0.834 (95%CI: 0.627-0.931, P < 0.001), respectively. Using the METAVIR fibrosis scoring system, 138 patients were scored as F1-F3 (47 HBV⁺, 91 HCV⁺): 62 F1, 56 F2, 20 F3 and 25 as F4 (14 HBV⁺ and 11 HCV⁺). A significant correlation was obtained between LSM and SSM (r² = 0.574, P < 0.0001); AUROC 0.839 (95%CI: 0.780-0.898) for METAVIR F1 vs F2-4, 0.936 (95%CI: 0.898-0.975) for F1-2 vs F3-4 and 0.932 (95%CI: 0.893-0.971) for F1-3 vs F4, all P < 0.001.

Considering the role of SS in predicting complications, Ye et al[69] evaluated the performance of liver and spleen stiffness, by ARFI elastography, for the assessment of liver fibrosis and esophageal varices in 66 patients with CHB, who underwent liver biopsy, and 138 with cirrhosis HBV related, examined by upper endoscopy. They obtained a significant linear correlation between liver (ρ = 0.87; P < 0.001) and spleen (ρ = 0.76; P < 0.001) stiffness and liver fibrosis. A fairly significant linear correlation (ρ = 0.65; P < 0.001) was found between SS and the varices grade, promising an important role of this technique also in HVB related cirrhosis, although further studies are required.

Interestingly, almost all the mentioned bio-chemical markers have been evaluated in combination with one another or with elastographic techniques in order to build sequential algorithms, aiming at correctly allocating and staging CHC and CHB patients. In CHB settings, Sebastiani et al[85] evaluated a series of possible algorithms combining various non-invasive markers concluding that the sequential combination of APRI, FibroTest and liver biopsy turned out the most promising approach.