Ultrasound elastography: liver

Abstract

Ultrasound elastography, also termed sonoelastography, is being used increasingly in clinical practice to aid the diagnosis and management of diffuse liver disease. Elastography has been shown to be capable of differentiating advanced and early-stage liver fibrosis, and consequently a major application in clinical liver care includes progression to cirrhosis risk stratification through (1) assessment of liver fibrosis stage in HCV and HBV patients, (2) distinguishing non-alcoholic steatohepatitis from simple steatosis in non-alcoholic fatty liver disease patients, and (3) prognostic evaluation of liver disease is autoimmune liver disease. In addition, elastographic characterization of focal liver lesions and evaluation of clinically significant portal hypertension have the potential to be clinically useful and are areas of active clinical research.

Ultrasound elastography, or the estimation of tissue stiffness by ultrasound, refers to a group of technologies that estimate tissue stiffness, including real-time/static elastography (RTE), transient elastography (TE), acoustic radiation force impulse (ARFI) imaging and real-time shear wave elastography (SWE). A comprehensive discussion of the advantages and disadvantages of these technologies is beyond the scope of this article. However, a conceptual overview is helpful to the practicing clinician. Broadly, ultrasound elastography techniques can be divided into non-imaging and imaging techniques, and the imaging techniques can be further subdivided into strain imaging and shear wave imaging techniques.

Non-imaging techniques

TE is an ultrasound-based non-imaging technique in which a mechanical actuator vibrates on the skin, inducing the propagation of low frequency mechanical waves through the underlying liver. The velocity of these waves is related to liver stiffness, and can be measured with ultrasound, yielding a quantitative liver stiffness estimate, measured as the hepatic Young’s modulus in kilopascals (kPa). This technique does not produce anatomic images.

Imaging techniques

Strain elastographic techniques rely on tissue deformation induced by operator pressure, patient body motion, or arterial pulsation. Tissue deformation measures are sonographically derived and mapped onto the anatomy. These “deformation images” overlie the conventional B-mode image, and allow assessment of which portions of the imaged tissue are relatively soft or hard. Strain imaging is commonly referred to as “semi-quantitative elastography” as it is unable, in the absence of a known deformation force, to yield absolute estimates of tissue stiffness.

Shear wave techniques, also termed ARFI techniques or SWE techniques, employ acoustic radiation force to induce microscopic tissue movements, producing tissue shear waves. In these techniques, both deformation force and tissue deformation are known, and quantitative estimates of tissue stiffness, expressed as Young’s modulus or shear wave velocity can be obtained.

In this paper, we review applications of ultrasound elastographic technologies in imaging of the liver; evaluation of diffuse liver disease, characterization of focal liver lesions, and the evaluation of portal hypertension.

Diffuse liver disease: overview

Diffuse liver disease is the commonest disease in humans, with an estimated US prevalence of 14.78% [1]. In 2010, cirrhosis resulted in an estimated 49,500 US deaths, and more than one million deaths worldwide, accounting for 1.95% of all global deaths [2, 3].

The etiologies of diffuse liver disease are numerous, and include hepatitis viral disease, alcoholic liver disease, non-alcoholic fatty liver disease (NAFLD) , and autoimmune hepatitis [4, 5]. Irrespective of etiology, chronic diffuse liver disease follows a common pathophysiologic pathway in which repeated episodes of liver injury are followed by healing, regeneration, and fibrosis, ultimately culminating in an irreversible state of fibrosis-induced hepatic dysfunction termed cirrhosis (Fig. 1) [6]. Cirrhosis is a severe and costly disease state (Fig. 2).

Fig. 1.

Strain elastography in a healthy liver (A) and in a liver with moderate fibrosis (B). *Images courtesy of Hitachi Aloka Medical.

Fig. 2.

Pathophysiology of liver fibrosis, if left untreated it progresses to cirrhosis irrespective of the etiology of chronic liver disease. Reprinted from [6].

Hence, irrespective of etiology, the goal of management in early diffuse liver disease is to prevent progression to cirrhosis. Reliable liver fibrosis staging is necessary to meet this goal. A number of staging systems have been proposed for liver fibrosis staging. The two commonest uses for the staging of fibrosis secondary to viral hepatitis are the METAVIR [7, 8] and Ishak [9] systems. The majority of the sonoelastography literature refers to the METAVIR liver fibrosis staging system, and consequently, this system is described in this review. In the METAVIR system, fibrosis is staged from Stage F0 to Stage F4, where Stage F0 is normal liver, Stage F1 is portal fibrosis without septa formation, Stage F2 is enlargement of portal tracts with rare septa formation, Stage F3 is numerous septa formation, and Stage F4 denotes cirrhosis in the form of nodular degeneration [7, 8]. The Scheuer scoring system (S0-S4) has been used in some elastography studies and has five Stages of fibrosis that show strong concordance (p < 0.001) with those of the METAVIR system [10].

Currently, the accepted liver fibrosis staging reference standard is histopathologic evaluation of non-focal liver biopsy specimens. However, liver biopsy has a number of limitations: (1) liver biopsy is invasive, with an estimated mortality rate of 0%–0.13%, a post-procedure pain rate of up to 63.9%, and patient anxiety [11–14], (2) liver biopsy is costly, (3) inter-observer variability limits the clinical interpretability of liver biopsy samples [15, 16], and (4) sampling error reduces accuracy—liver biopsy samples only 1/50,000th of the liver, and fibrosis is typically heterogeneously distributed [17, 18].

Owing to these limitations, a number of non-invasive liver fibrosis staging alternatives have been proposed, including serum tests, magnetic resonance elastography (MRE), and ultrasound elastography. Serum biomarkers have been shown to have varying accuracy of 0.78–0.80 for diagnosis of significant fibrosis (≥F2) [19–26], and may be confounded by other disease processes, while MRE has been shown to be accurate [27], but is presently costly. It is likely that all of these non-invasive modalities—blood testing, ultrasound elastography, and MRE—will ultimately play a role in diffuse liver disease care. In this review, we focus on the role of ultrasound elastography in several distinct diffuse liver diseases.

Hepatitis C viral (HCV) disease

Hepatitis C viral disease has an estimated prevalence of 5.2 million in the United States [28] and >185 million worldwide [29]. Hepatitis C disease is typically asymptomatic, and progresses insidiously, commonly presenting decades after initial infection with irreversible cirrhosis [30, 31]. While HCV is presently the commonest cause of cirrhosis in the US, only 15%–30% of HCV-infected patients progress to cirrhosis over a 20–30 year period, with the remaining 70%–85% having disease that progresses more slowly or not at all [32–34]. Recent HCV therapy advances have led to the emergence of molecularly targeted therapies, capable of curing HCV [35]. However, these treatments are costly, with a per patient drug expenditure of up to $80,000 [35]. If every HCV-infected patient in the US were treated, the drug costs alone would be an unaffordable ~$400 billion. It is therefore optimal that HCV treatment be allocated to the 15%–30% of HCV-infected patients at risk for developing cirrhosis.

In prior studies, it has been shown that liver fibrosis of stage ≥F2 was predictive of subsequent cirrhosis in HCV-infected patients [36]. As a result, the primary mechanism by which HCV patients are risk-stratified for cirrhosis is through liver biopsy. If liver biopsy shows fibrosis ≥F2 and does not show cirrhosis, which is, by definition, irreversible, the patient would optimally be treated to prevent cirrhosis. However, this care paradigm requires the performance of liver biopsy in all HCV patients. With an estimated 5.2 million HCV-infected patients in the USA, cost, morbidity, and complexity render liver biopsy unsuitable for widespread use for patient selection for therapy [28]. TE has shown a promise with intermediate to high diagnostic accuracy in diagnosing HCV patients with ≥F2 liver fibrosis [37–43]. At varying cut-off values of 6.2–8.7 kPa in different studies, TE has shown an area under the receiver operating characteristic (ROC) curve of 0.77–0.90 for assessment of significant fibrosis (F ≥ 2) [44]. Shear wave elastographic approaches have both shown comparable or better accuracy than TE for the diagnosis of ≥F2 liver fibrosis in HCV patients [45–48]. Figure 3 shows SWE images in patients with F0 and F3 stage fibrosis, respectively.

Fig. 3.

Panel A: A SWE elastogram of a 65-year-old male with chronic hepatitis C with Stage F0 fibrosis on pathological examination shows an estimated liver Young’s modulus of 5.9 kPa. Panel B: A SWE elastogram in a 60-year-old male with chronic hepatitis C with stage F3 fibrosis on pathological examination shows an estimated liver Young’s modulus of 13.4 kPa. The B-mode images are indistinguishable.

Figure 4 shows a box and whisker plot of liver stiffness measurements using ARFI-based SWE for varying fibrosis stages as seen on liver biopsy. These values as obtained from 911 patients with Hepatitis C in a large multi-center trial clearly indicated a strong correlation between liver stiffness and fibrosis (r = 0.654, p < 0.0001) [49]. In the study, Sporea et al. [49] reported a sensitivity and specificity of 69.1% and 79.8%, respectively, with an area under the ROC curve of 0.792 to differentiate ≥F2 fibrosis at a cut-off value of 1.33 m/sec.

Fig. 4.

Box and whisker plot showing liver stiffness values for different liver fibrosis stages as seen on liver biopsy. The study included 911 patients in a large multi-center trial using ARFI shear wave elastography. Reprinted from [49].

Similarly, both TE and SWE have shown high accuracy for the diagnosis of cirrhosis (F4) [45–47, 50]. In their multi-center study, Sporea et al. reported a sensitivity and specificity of 84.3% and 76.3%, respectively, at a cut-off value of >1.55 m/sec to diagnose cirrhosis with an area under the ROC curve of 0.842.

In a cost-effectiveness study performed in the UK, Canavan et al. [51] concluded that annual TE is an ideal strategy to diagnose cirrhosis in patients with HCV, potentially reducing the number of liver biopsies performed per year in the UK by 2,000 and increasing the diagnosis of cirrhosis by 20%. The additional cost of this strategy was approximately $160 (£98.78) per patient, and its yield was an additional 1.72 unadjusted life years per patient as compared with the current practice of liver biopsy [51]. This is well below most societal willingness-to-pay thresholds, and suggests this strategy could be used in many healthcare settings to meaningfully reduce liver disease morbidity and mortality.

Hepatitis B viral (HBV) disease

HBV is one of the most common infections in humans, with an estimated 2 billion infected people worldwide [52]. The majority of infected people clear the virus. A minority remains chronically infected and can develop cirrhosis secondary to chronic liver inflammation. This minority comprises over 350 million people globally [52, 53]. The spectrum of disease in those infected with HBV varies from inactive carrier state to progressive chronic hepatitis B. Similar to HCV, the finding of ≥F2 fibrosis on liver biopsy carries an increased risk of subsequent cirrhosis [54].

Inactive HBV carriers have been shown to have liver stiffness values similar to healthy individuals, as measured by TE. Oliveria et al. [55] reported a mean LS value of 5.0 ± 1.8 kPa in 68 subjects with inactive HBV, values very similar to those reported by Maimone et al. [56] who reported a value of 4.8 ± 1.2 kPa in 125 inactive carriers. In a meta-analysis, Branchi et al. [53] reported that TE has a sensitivity and specificity of 70%–93% and 38%–92%, respectively, for the diagnosis of significant fibrosis (≥F2) with cut-off values ranging from 5.2–8.7 kPa in patients with chronic HBV [55, 57–62].

Friedrich-Rust et al. [63] performed a study in patients with HBV using ARFI and TE and found no significant difference between the two methods with ARFI and TE having a diagnostic accuracy as expressed by the areas under the ROC curves of 0.75 and 0.83, respectively, to diagnose significant fibrosis (≥F2). Zhang et al. [64] reported similar results with the areas under the ROC curve for diagnosing ≥Stage 2 fibrosis (Scheuer scoring system) of 0.764 and 0.813 for ARFI and TE, respectively.

Non-alcoholic fatty liver disease (NAFLD)

Non-alcoholic fatty liver disease (NAFLD) is a major public health problem. It is the commonest United States liver disorder, with a prevalence of 17%–46% [65] and a worldwide prevalence of 6%–35% with a median of 20% [66]. NAFLD has two forms: (1) Simple steatosis (SS, 80% of cases), defined as excess liver fat without inflammatory injury, and (2) Non-alcoholic steatohepatitis (NASH, 20% of cases), in which excess liver fat is associated with inflammation, which results in fibrosis and regeneration, culminating in cirrhosis [66]. NAFLD presents a major challenge: it is clinically essential to accurately distinguish the 20% with NASH at risk for cirrhosis from the 80% with SS who will not progress and require no treatment. CT or MRI can quantify liver fat, but CT uses ionizing radiation, and MRI is costly. Moreover, conventional CT and MRI cannot distinguish SS and NASH, which currently requires non-targeted liver core biopsy. It is impractical for 17–46% of the population to undergo costly and invasive liver biopsy.

The majority of studies examining whether hepatic steatosis confounds liver stiffness assessment have concluded that it does not significantly affect liver stiffness [46, 47, 63, 68–71]. It is, however, well established that hepatic inflammation increases estimated liver stiffness, potentially confounding fibrosis staging [49, 69, 72–77]. As a result, liver stiffness in NASH may be increased as a result of fibrosis and/or inflammation (Fig. 5). It is presently unclear whether SWE can detect early-stage NASH, and which quantitative estimate of Young’s modulus might be an optimal cut-off level for differentiating NASH and SS. Nonetheless, the limited available evidence strongly suggests ultrasound elastography is likely to be a valuable test for the differentiation of NASH and SS.

Fig. 5.

Panel A: A SWE elastogram of a 35-year-old male patient with Grade 4 steatosis but no inflammation or fibrosis shows an estimated Young’s modulus of 5.5 kPa, below the cut-off for the diagnosis of stage F1 fibrosis. Panel B: A SWE elastogram depicts a SWE Young’s modulus estimate of 8.5 kPa in a patient that has the same stage of fibrosis (Stage F0), lesser steatosis (grade 3) but more inflammation as characterized by a total activity score of 1/3 on METAVIR grading after liver biopsy. The B-mode images are indistinguishable.

Transient elastography has shown areas under the ROC curves of 0.78–0.99 using variable cut-off values (6.6–7.8 kPa) to differentiate significant fibrosis (≥F2) in several clinical studies [78–83] with sensitivity and specificity of 67%–88% and 64%–91%, respectively. Similarly, ARFI-based SWE has shown an area under the ROC curve of 0.944 [76] in differentiating ≥F2 fibrosis and other studies have shown AUROC values of 0.90–0.97 in differentiating ≥F3 fibrosis in patients with NAFLD [67, 84]. Furthermore, a study by Guzman-Aroca et al. [74] showed that NAFLD could be differentiated from patients with NASH or fibrosis using ARFI elastography with an accuracy of 0.899 with a sensitivity of 85% and specificity of 83.3% at an optimal cut-off value of 1.3 m/sec. Yoshioka et al. [85] provide a detailed summary of studies evaluating liver stiffness measurement in patients with NAFLD.

Alcoholic liver disease (ALD)

The prevalence of alcoholic liver disease in the United States is as high as 2.05%, making it the second most common cause of CLD in the United States [1]. Although alcoholism is associated with a number of diseases, alcoholic liver disease remains the highest cause of mortality in these patients with ~50% of cirrhosis associated deaths in 2011 in the United States being attributed to alcoholic liver disease [86]. Similar to other causes of CLD, patients with ALD require the precise estimation of liver fibrosis in order to monitor progression and treatment.

Similar to NASH, alcoholic liver disease has histologic findings of steatosis and inflammation resulting in fibrosis. If untreated, ALD may progress to cirrhosis [87]. Nahon et al. [88] showed that liver stiffness measurements using TE correlated with fibrosis stage (r = 0.70, p < 0.0001) in 147 patients with alcoholic liver disease. They showed that the areas under the ROC curves were 0.94 and 0.87 for the diagnosis of extensive fibrosis (≥3) and cirrhosis as per the Brunt scoring system. Nguyen-Khac et al. [89] found that TE was better than seven non-invasive laboratory tests in the assessment of liver fibrosis with a correlation of 0.72 (p < 0.014) in 103 patients with alcoholic liver disease. Areas under the ROC curves to diagnose significant fibrosis (≥F2) and cirrhosis were 0.91 and 0.92, respectively.

Zhang et al. [90] showed a strong correlation (r = 0.685, p < 0.001) between histological fibrosis and liver stiffness in patients with ALD, as measured by ARFI elastography. Using the Scheuer scoring system for histologic grading of fibrosis, they found areas under the receiver operating characteristic curve of 0.846 and 0.893 for the differentiation of significant fibrosis (≥S2) and cirrhosis (S4), respectively [90].

Bardou-Jacquet et al. [91] showed that alcohol abstinence can significantly decrease liver stiffness values as measured by TE, implying a potential role for liver elastography in providing an objective metric of compliance with abstinence regimes.

Autoimmune liver disease (Autoimmune hepatitis, primary sclerosing cholangitis, and primary biliary cirrhosis)

Primary biliary cirrhosis (PBC).

PBC is usually diagnosed clinically, with supportive evidence provided by antibody testing. Histopathologic examination is currently performed at the time of diagnosis primarily to provide a baseline for evaluation of treatment response and also for prognostic evaluation [92]. Corpechot et al. [93] demonstrated that liver stiffness measured with TE shows a high correlation with fibrosis stage in patients with PBS and PSC. Similarly, Zhang et al. [94] demonstrated PBC could be accurately staged with SWE. Paradoxically, PBC treatment with methotrexate can cause liver fibrosis [95]. Consequently, PBC patients on long-term methotrexate are typically followed with sequential liver biopsy. This cohort of PBC patients is likely to benefit substantially from elastographic liver fibrosis monitoring.

Primary sclerosing cholangitis.

The overall incidence of PSC in the United States, adjusted for age and sex, is estimated at 0.9 per 100,000 population, with a prevalence of 13.6 per 100,000 [96]. Similar to PBC, PSC often requires histological examination to confirm diagnosis and for purposes of staging. Corpechot et al. [93] demonstrated a strong correlation (r = 0.84, p < 0.001) of histologic fibrosis stage and liver stiffness measured by TE. In a longitudinal follow-up study [97], they concluded that TE provides a high diagnostic accuracy for ≥F2 fibrosis with a sensitivity and specificity of 72% and 89%, respectively, at a cut-off value of 8.6 kPa and an area under the ROC curve of 0.84. They also concluded that baseline measurements and longitudinal changes serve as prognostic factors in PSC follow-up. In an independent cohort of 168 PSC patients, they found that a liver stiffness measurement >11.1 kPa or an increase of1.5 kPa over a 1-year interval was associated with ten times higher adverse outcomes risk (death, liver transplantation, or hepatic complications) within a 4-year period [97]. However, as correctly pointed out by Ehlken et al. [98], it is important to understand that several confounders, such as food intake, cholestasis, inflammation, and acute hepatitis can result in higher liver stiffness measurements and are not necessarily indicative of worsening chronic disease. Cholestasis due to biliary obstruction is present in as many as 40%–50% PSC patients and should be considered when interpreting elastography measurements.

Autoimmune hepatitis (AIH).

AIH is a diagnosis of exclusion. Even though biopsy may be necessary at the time of diagnosis, progression can be monitored using elastography. There is limited literature regarding the evaluation of elastography and autoimmune hepatitis. However, Righi et al. [99] have demonstrated that SWE can be used in these patients to differentiate higher grades of fibrosis from healthy volunteers. Extrapolating cut-off values from other disease states to AIH should be done with caution, however, as Romanque et al. have reported that autoimmune hepatitis produces disproportionately higher liver stiffness values [100].

Focal liver lesions

Conventional B-mode ultrasound has been shown to be cost-effective for screening for focal liver lesions in cirrhotic patients, but has lower specificity than contrast-enhanced US, CT, or MR for the characterization of detected lesions [101]. However, the administration of intravenous contrast requires intravenous access, and is more costly and time-consuming than conventional non-contrast ultrasound. Sonoelastography offers a potential additional non-contrast ultrasound biomarker to differentiate benign and malignant liver lesions. All sonoelastography technologies presently have limitations for the assessment of focal liver lesions: (1) TE is not an imaging test and therefore cannot be used for focal lesions, (2) strain (compression) elastography lacks the ability to evaluate lesions at greater depths and does not provide true quantitative estimates of lesion stiffness, and (3) SWE acoustic pulses typically do not yield interpretable shear wave measurements more than 6–8 cm deep to the skin surface.

A 2013 meta-analysis of the ability of RTE and ARFI elastography to distinguishing malignant from benign liver lesions showed overall sensitivity and specificity of 85% (95% CI 80%–89%) and 84% (95% CI 80%–88%), respectively [102]. However, the subsequent literature on this subject is inconsistent. While Cho et al. [103] showed that ARFI elastography could differentiate malignant lesions with a sensitivity and specificity of 89% and 81%, respectively, at a shear wave velocity cut-off of 2 m/s, Heide et al. [104] concluded that there was no statistically significant difference between elasticity values of malignant and benign focal liver lesions. Similarly Park et al. [105] concluded that while elastography provides additional information for the characterization of focal liver lesions, there is significant overlap, preventing differentiation of malignant from benign lesions based on elasticity values. Ronot et el. [106] came to a similar conclusion regarding SWE for focal liver lesion characterization. Guibal et al. [107] similarly showed a significant overlap between hemangiomas and hepatocellular carcinoma, with mean elastographically estimated Young’s modulus of 13.8 ± 5.5 and 14.86 ± 10 kPa, respectively.

In Table 1 (SWE values in m/sec) and Table 2 (SWE values in kPa) a significant overlap in estimates of lesion stiffness is evident across benign and malignant liver lesions.

Table 1.

Summary of ARFI values in m/sec for all different types of lesions

Studies	Final diagnosis based on	Metastasis	HA	HCC	ccc	Adenoma	FNH
Cho et al. [103]	Biopsy and Imaging	NSS	1.51 ± 0.71 (n = 11)	2.45 ± 0.81 (n = 17)	NSS	NS	NS
Davies et al. [108]	Biopsy and Imaging	4.23 ± 0.59 (n= 10)	1.35 ± 0.48 (n = 35)	NS	NS	NS	NS
Gallotti et al. [109]	Surgery, biopsy, and imaging	2.87 ±1.13 (n = 9)	2.30 ± 0.95 (n = 7)	2.17 ± 0.85 (n = 6)	NS	1.25 ± 0.37 (n = 5)	2.75 ± 0.95 (n = 13)
Kim et al. [110]	Surgery. Biopsy, and imaging	CCM: 3.70 ± 0.61 (n = 20) OM: 2.82 ± 0.96 (n = 24)	1.80 ± 0.57 (n = 28)	2.66 ± 0.94 (n = 26)	3.27 ± 0.64 (n = 3)	NS	NS
Park et al. [105],	Biopsy and imaging	2.35 ± 1.18 (n = 8)	1.83 ± 0.62 (n = 5)	2.48 ± 0.84 (n = 24)	1.65 ± 1.43 (n = 7)	NS	0.97 ± 0.48 (n = 3)
Zhang et al. [Ill]	Surgery, biopsy, and imaging	3.20 ± 0.62 (n = 39)	1.33 ± 0.38 (n = 28)	2.59 ± 0.91 (n = 61)	3.74 ± 0.54 (n = 12)	NS	1.90 ± 0.45 (n = 14)
Yu et al. [112],	Biopsy and imaging	2.73 ± 0.89 (n = 13)	1.75 ± 0.8 (n = 35)	2.49 ± 1.07 (n = 28)	NS	1.79 ± 0.14 (n = 2)	2.18 ± 0.84 (n = 15)

HA, hemangioma; HCC, hepatocellular carcinoma; CCC, cholangiocarcinoma; FNH, focal nodular hyperplasia; NS, not studied; NSS, not studied separately; CCM colon cancer metastases; OM other metastases

Table 2.

Summary of ARFI values in kPa for all different types of lesions

Studies	Final diagnosis based on	Metastasis	Focal fatty sparing	HA	HCC	ccc	Adenoma	FNH
Ronot et al. [106]	Surgery, biopsy, and imaging	NS	11.3 ± 4.3 (n = 5)	17.1 ± 7 (n = 20)	19.6 (n = 1)	34.1 ± 7.3 (n = 2)	19.7 ± 9.8 (17)	33.3 ± 12.7 (60)
Guibal et al. [107]	Biopsy and Imaging	28.8 ± 16 (n = 53)	6.6 ± 0.3 (n = 3)	13.8 ± 5.5 (n = 22)	14.86 ± 10 (n = 26)	56.9 ± 25.6 (n = 7)	9.4 ± 4.3 (n = 10)	33 ± 14.7 (n = 16)

HA, hemangioma; HCC, hepatocellular carcinoma; CCC, cholangiocarcinoma; FNH, focal nodular hyperplasia; NS, not studied

More research is needed before elastography can be used clinically to characterize focal liver lesions.

Portal hypertension (PH)

Portal hypertension is the hallmark of cirrhosis, and is best diagnosed by directly measuring the hepatic venous pressure gradient (HVPG) [113–115]. Cirrhosis can be staged into four categories (Table 3) where Stage 1 and Stage 2 represent compensated liver disease, and Stage 3 and 4 represent decompensated liver disease. The four stages of cirrhosis are demarcated by clinical factors and HVPG measurements (Table 3). In a large systematic review of 108 studies, it was found that the median 1-year survival in the four stages of cirrhosis was 99%, 97%, 80%, and 43%, respectively [114, 116].

Table 3.

Stages of cirrhosis [117, 118]

		Ascites	Varices	HVPG (mm Hg)
Compensated liver disease	Stage 1	No	No	6–10
	Stage 2	No	Present	10–12
Decompensated liver disease	Stage 3	Present	Present or absent	>12
	Stage 4	Present or absent	Variceal bleeding

HVPG is the most reliable predictor of varices and ascites development [117]. Clinically significant portal hypertension is typically defined as the development of varices and complications of portal hypertension, including ascites, variceal bleeding, or hepatic encephalopathy. These complications generally occur at HVPG values greater than 10 mm Hg [119]. However, measuring HVPG is invasive, requires expertise, and costs ~$4,000 in the United States [113].

Liver and spleen stiffness as portal hypertension biomarkers have been studied in the last few years. In a meta-analysis, Shi et al. [120] reported a sensitivity and specificity of TE liver stiffness measurements for the detection of significant portal hypertension to be 0.90 (95% CI 0.81–0.95) and 0.79 (95% CI 0.58–0.91), respectively. Colecchia et al. showed that spleen stiffness measurements are also good predictors of clinically significant portal hypertension using TE (AUROC = 0.966). A recent study by Elkrief et al. [121] compared TE and SWE for the prediction of portal hypertension in 61 patients who had HVPG measurements. They reported better technical success and diagnostic accuracy for SWE for diagnosing clinically significant portal hypertension (>10 mm Hg) as compared to TE, in both the liver (0.87 vs. 0.78) and spleen (0.64 vs. 0.63). A liver stiffness value of 24.6 kPa with SWE had sensitivity, specificity, and accuracy of 81%, 88%, and 82%, respectively, for diagnosing clinically significant PH, as compared to 52%, 100%, and 57%, respectively, for a TE value of 65.3 kPa.

Summary

Ultrasound elastography is an established liver fibrosis biomarker, with the potential to replace many non-focal liver biopsies presently performed as part of routine clinical care. A number of confounding factors exist, including the effects of inflammation, meals, cholestasis, and variation in acquisition technique. As a result, effective interpretation of liver elastography requires expertise and careful review of patient history. Abdominal imagers should recognize that stiffness cut-offs do not necessarily generalize from one disease to another, from one measurement device to another, and from one ethnic group to another [49]. While there is evidence that ultrasound elastography can be used to non-invasively diagnose portal hypertension, the role of ultrasound elastography for the detection and characterization of focal liver lesions has not yet been established.[111]