State of the Art Imaging for Chronic Liver Disease

Khaled M. Elsayes Professor of Diagnostic Radiology University of Texas, MD Anderson Cancer Center and McGovern Medical School

Chronic liver disease is an increasingly common and important disorder, now afflicting more than 25% of the world’s population. Regardless of its underlying cause, chronic liver disease can progress to cirrhosis, development of liver cancer, and liver-related death. Recent advances in imaging knowledge and technology have elevated the role of radiology in the diagnosis and management of patients with or at risk for chronic liver disease across its entire spectrum and course.

Noninvasive Monitoring of Chronic Liver Disease

With a rising epidemic of obesity and metabolic syndrome in the United States, nonalcoholic fatty liver disease and its progressive form, nonalcoholic steatohepatitis, are increasingly encountered in practice. Additionally, overconsumption of alcohol and viral hepatitis remain common causes of chronic liver disease. Irrespective of etiology, chronic liver disease represents a spectrum of inflammation, injury, fibrosis, and eventually, cirrhosis. Although in the past, cirrhosis was considered an irreversible injury, recent medical advances provide opportunities to halt and even reverse fibrosis. Likewise, identification and modulation of risk factors, like fat and iron deposition in the liver, are essential to managing patients at risk for or with chronic liver disease. Pathological assessment remains a reference standard, but risks and cost associated with biopsy reduce the benefit in the setting of longitudinal monitoring. Radiologists play an essential role in providing noninvasive quantitative information to direct management.

State of the art imaging options for assessing chronic liver disease focus on three key elements: liver fat, iron, and fibrosis. Liver fat can be identified by ultrasound, CT, and MRI. The most accurate, accessible, and precise method for monitoring patients with chronic liver disease is MRI-proton density fat fraction (PDFF). MRI-PDFF takes advantage of the fact that owing to differences in molecular structures, fat and water protons experience different magnetic fields and precess at different rates. This offset in frequency (“chemical shift”) is the basis for in- and opposed-phase dual echo imaging with its familiar appearance, including “India ink” etching at fat-water interfaces on opposed-phase images, due to signal cancellation when water and fat are precessing directly opposite each other. MRI-PDFF expands on this concept and is estimated by acquiring images at multiple echo times, selected to optimize separation of fat and water signals, and by taking into account several confounders that otherwise introduce errors into fat quantification. Iron is one of the most important confounders and is also an important factor contributing to chronic liver disease. Iron is ferromagnetic and causes signal decay (T2*), due to disruption of local magnetic fields. The decay of signal over time (R2*) is directly proportional to iron content over a wide pathophysiological spectrum, allowing us to estimate R2* values and convert them to liver iron concentrations. Hence, MRI-PDFF quantifies both fat and iron.

The most validated and clinically used method for estimating liver fibrosis is elastography. Elastography can be done with ultrasound or MRI. Elastography is an imaging technique that quantifies the stiffness of tissue, or resistance to deformation following application of external pressure. Imaging methods estimate stiffness by generating shear waves in the liver and measuring their propagation. Ultrasound methods differ based on how they generate shear waves and whether they produce gray-scale images (point shear wave elastography and 2D elastography) or not (vibration controlled transient elastography). Ultrasound methods measure shear wave speed, which can be converted into tissue stiffness values. The speed measurements may differ between manufacturers and etiologies of chronic liver disease, which challenges establishment of universal thresholds for stages of fibrosis. Magnetic resonance elastography (MRE) utilizes a standard system for shear wave generation and measurement across all vendors and platforms; as a result, the tissue stiffness values obtained from MRE are thought to be more reproducible.

Standardizing Diagnosis of Liver Cancer in Patients with Chronic Liver Disease

Hepatocellular carcinoma (HCC) can be confidently diagnosed based on imaging, in contradistinction to most malignancies that require tissue examination for their diagnosis. The noninvasive diagnosis of HCC is justified by the high positive predictive value of CT and MRI for this purpose when stringent criteria are applied in high-risk patients (i.e., high pre-test probability) [1–3]. Additionally, cross-sectional imaging assesses local spread and distant metastases.

Current standards in the noninvasive diagnosis of HCC follow the guidelines of the American Association for the Study of Liver Disease (AASLD), Organ Procurement and Transplantation Network (OPTN), and Liver Imaging Reporting and Data System (LI-RADS) [4–6]. These guidelines agree on certain imaging features that should be present in an observation to provide the required high positive predictive value for HCC, such as a maximum diameter of at least 10 mm and characteristic dynamic enhancement characteristics [5, 6] discussed further below. Distinct differences used to exist between the three guidelines in the categorization of hepatic lesions, until the release of the latest LI-RADS guidelines for CT and MRI in 2018. The latest release comprised minor modifications to LI-RADS version 2017 to facilitate its integration into the AASLD clinical practice guidelines in August 2018 [7]. LI-RADS and AASLD now have identical criteria for definite HCC, and the OPTN criteria are nearly identical to LI-RADS and AASLD.

LI-RADS is a comprehensive system that provides standards for terminology, technique, interpretation, and reporting of liver imaging. It has been developed by a multi-disciplinary and increasingly international team of diagnostic and interventional radiologists, hepatobiliary surgeons, hepatologists, and hepatopatholgists, alongside support from the American College of Radiology [8]. Since its first release in 2011, LI-RADS has been updated periodically, with the latest update in 2018 [9–11]. LI-RADS assigns a diagnostic category code for each observation to communicate the likelihood of being benign or being HCC, ranging from LR-1 (definitely benign) to LR-5 (definitely HCC). The LR-5 category has a reported specificity of 95% for HCC [6]. In addition to the previous five categories, LI-RADS also provides three other categories—LR-NC (not categorizable), LR-TIV (tumor in vein), and LR-M (probably or definitely malignant, not necessarily HCC)—with certain criteria for each category.

The imaging diagnosis of HCC in LI-RADS is based on the presence or absence of five major imaging features and a number of ancillary features (AFs). Major features include nonrim arterial phase hyperenhancement (APHE), nonperipheral “washout” appearance, enhancing “capsule” appearance, size, and threshold growth.

The AFs are divided into three groups: AFs that favor malignancy in general, AFs that favor HCC in particular, and AFs that favor benignity. A preliminary LI-RADS category is assigned based on the present major features, then the AFs are used at the interpreter’s discretion to adjust the preliminary category [6].

In summary, the latest advances in imaging of HCC allow for a confident noninvasive diagnosis of this malignancy and comprehensive assessment of other lesions and pseudolesions depicted by imaging.

Improving Sensitivity for Liver Cancer Diagnosis with HBAs

HBAs are gadolinium-based intravenous MR contrast agents that permit hepatobiliary phase (HBP) imaging, in addition to conventional dynamic post-contrast phases. Gadoxetate disodium is the most commonly utilized HBA, due to high hepatobiliary excretion and convenient HBP timing of 10–30 minutes [1].

Gadoxetate offers several advantages for patients with cirrhosis. Of all available modalities, gadoxetate-enhanced MRI has the highest overall per-lesion sensitivity (86%) and positive predictive value (94%) for diagnosis of HCC, as well as the highest sensitivity (84–96%) for detection of ≤ 2 cm HCCs [2–6]. Unlike APHE, reduced gadoxetate uptake is an early event in hepatocarcinogenesis: up to 38% of early HCCs may be seen only on the HBP, and 82% of high-grade dysplastic nodules (DN) and 76% of early HCCs are hypointense on the HBP [7–10].

HBP hypointense nodules without APHE are unique to HBA MRI. If sampled histologically, 74% of such nodules are HCCs, and 10% are DN, although these numbers may be inflated by selection bias; if followed, 16–43% progress to hypervascular HCC within 24 months [11–18]. In patients who undergo resection for early-stage HCC, the presence of HBP hypointense nodules predicts high HCC recurrence risk and lower overall survival [19]. Intermediate to long-term recurrence-free survival may be improved, if these nodules are treated concomitantly at the time of HCC resection [20].

HBP hypointense nodules without APHE are also markers of increased HCC risk elsewhere in the liver: the cumulative three year rate of HCC elsewhere in the liver is 22%, compared to 6% in patients with no such nodules [21]. Presence of these nodules in patients with early HCC is associated with decreased recurrence- free survival and higher intrahepatic recurrence rates following resection or ablation [22–24].

The degree of gadoxetate uptake may predict tumor differentiation: poorly-differentiated HCCs are more frequently HBP hypointense (98%), compared with well- or moderately-differentiated HCCs (86%) [25]. Up to 15% of HCCs may be iso- or hyperintense on the HBP, and such HCCs have more favorable outcomes, including improved recurrence-free and overall survival [26–28].

Use of gadoxetate in patients with cirrhosis is associated with several important pitfalls. Smaller contrast dose and volume affect timing of the arterial phase (AP) and may lead to reduced peak enhancement of HCC in the AP [29, 30]. Furthermore, gadoxetate is associated with higher incidence (5–22%) of transient severe motion, which occurs at or around the time of the late AP and leads to image degradation [31–33]. Poor quality of the AP may affect depiction of APHE, a feature that is required for noninvasive HCC diagnosis [34].

Portal venous phase (PVP) “washout” appearance in combination with APHE allows for nearly 100% specificity of HCC diagnosis [35]. Parenchymal uptake of gadoxetate starting as early as the PVP results in observations potentially appearing relatively hypointense to the parenchyma, due to lower uptake of gadoxetate rather than true “washout.” As a result, hypointensity in the transitional phase (TP) is not equivalent to hypointensity during the PVP or delayed phases with extracellular agents: if hypointensity in the TP is considered “washout,” the specificity for HCC decreases from 98–100% to 86–95% [36]. Therefore, LI-RADS restricts assessment of “washout” with gadoxetate to the PVP [37]. Another effect of the early parenchymal enhancement with gadoxetate is the potential to obscure enhancement of the “capsule.”

In patients with decompensated cirrhosis, diminished parenchymal uptake of gadoxetate results in less enhancement during the TP and the HBP [38]. As a result, conspicuity of HCC in the HBP is decreased in patients with poor hepatic function [39]. Additionally, interpretation of HBP intensity of liver observations— particularly if iso- or hyperintense to the background— may be unreliable in the setting of suboptimal HBP enhancement.

Although HBP hypointensity improves detection of HCC and high-grade DN, TP and HBP hypointensity are not specific to HCC, as any lesion without functional hepatocytes (e.g., cysts, hemangiomas, non-HCC malignancies, etc.) will appear hypointense in the HBP [40].

In conclusion, use of gadoxetate in patients with cirrhosis offers certain advantages—particularly higher sensitivity for HCC, if liver function is preserved and AP quality is adequate—but radiologists should be aware of the various pitfalls of gadoxetate to optimize patient selection and image interpretation.

Acknowledgments

Kathryn Fowler, Claude Sirlin, and Victoria Chernyak also contributed to this article.

References

---

The opinions expressed in InPractice magazine are those of the author(s); they do not necessarily reflect the viewpoint or position of the editors, reviewers, or publisher.