Clinical Significance of Liver MR Imaging

Abstract

MRI is widely used in clinical practice for detecting liver diseases. Since the introduction of gadoxetic acid, MRI has become the most effective modality for the detection and characterization of focal liver lesions. According to previous meta-analyses, the area under the receiver operating characteristic curve (AUROC) was 0.97–0.99 for the diagnosis of small hepatocellular carcinoma (≥ 2 cm) by gadoxetic-acid-enhanced MRI. Moreover, the AUROC for the diagnosis of colorectal liver metastases was significantly high (0.98). Despite gadoxetic acid’s drawbacks, its clinical utility outweighs them, making it the contrast agent of choice in routine liver MRIs. Moreover, clinically, liver MRI has become more prevalent for a quantitative assessment. Liver fibrosis can be evaluated using MR elastography; whereas, hepatic steatosis and iron overload can be evaluated using proton density fat fraction, with high accuracy and reproducibility. This article reviewed the usefulness of liver MRI, which can be a comprehensive imaging modality in clinical practice.

Introduction

The liver is the largest solid internal organ in humans and has many essential functions, especially in maintaining the internal environment, including detoxification, lipid and cholesterol homeostasis, bile production, and protein synthesis.¹ Despite damage to the liver, symptoms rarely appear; therefore, the liver is known as a “silent organ”. Consequently, it is important to detect abnormalities in the liver at an early stage and to intervene with appropriate treatment as soon as possible.

MRI is a well-established and widespread imaging modality for liver diseases that has improved significantly over the years with the advent of new hardware, software, and contrast agents. For diagnosing liver disease, MRI has comparatively more advantages than CT, including higher contrast resolution, lack of radiation exposure, and availability of two different types of contrast agents (i.e. extracellular and hepatocyte-specific contrast agents).² Furthermore, MRI provides morphological and quantitative information.

In recent years, CT and MRI have played an important role in the detection and characterization of focal liver lesions. Moreover, a hepatocyte-specific contrast agent (gadoxetic acid) has become widely used in routine clinical liver MRI due to its high performance in detecting and characterizing lesions.^3–5 As compared to contrast-enhanced CT, gadoxetic-acid-enhanced MRI has a better diagnostic ability, particularly for small lesions. Despite gadoxetic acid’s drawbacks, its clinical utility outweighs them, making it the contrast agent of choice in routine liver MRIs.

Chronic liver disease (CLD) is the 10th leading cause of death worldwide, with approximately two million deaths annually, and its incidence rate continues to rise.⁶ Viral hepatitis is the most common cause of CLD. Furthermore, nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH) are the most rapidly growing etiologies of CLD. Liver biopsy, followed by histopathological assessment, remains the gold standard for diagnostic evaluation of liver fibrosis, steatosis, and iron overload.⁷ According to a recent systematic review and meta-analysis, it is possible to safely perform a percutaneous ultrasound-guided liver biopsy with a low rate of major and minor complications;⁸ however, as a screening tool, it is difficult to conduct in large numbers because of its invasiveness and high cost. However, MRI is a noninvasive imaging tool that can evaluate liver fibrosis, steatosis, and iron overload in a single examination. Thus, this review discussed the clinical use of liver MRI in Japan and its clinical significance in the diagnosis of focal liver lesions, as well as the assessment of diffuse liver diseases.

Clinical Needs for Focal Liver Lesions

Detection of focal liver lesions

Recently, CT and MRI have played an important role in the accurate detection and characterization of focal liver lesions. CT is more accessible than MRI due to its shorter scan times and lower cost. However, MRI has more advantages over CT, such as better soft tissue contrast (Fig. 1) and no radiation exposure. Previously, CT arterial portography and CT hepatic arteriography were considered the tests with the highest diagnostic ability for liver lesions. However, since the introduction of gadoxetic acid, these tests are rarely performed solely for diagnostic purposes.

Fig. 1.

Multiple liver metastases from malignant melanoma in an 88-year-old man. (a) Unenhanced CT reveals several low-attenuation lesions in the liver (arrows). (b) Fat-saturated T2-weighted and (c) diffusion-weighted images reveal numerous hyperintense lesions in the liver. MRI has more advantages compared to CT, including better soft tissue contrast.

Currently, three different types of contrast agents—gadolinium-based (GBCAs), hepatocyte-specific (gadoxetic acid), and Kupffer cell-specific contrast agents (superparamagnetic iron oxide [SPIO])—are available for liver MRI. Fat-saturated T1-weighted images (T1WI) are obtained after GBCAs or gadoxetic acid injection; whereas, fat-saturated T2- or T2*-weighted images are taken after SPIO injection (Fig. 2).

Fig. 2.

Hepatocellular carcinoma in a 56-year-old man with liver cirrhosis due to hepatitis C virus infection. (a) Precontrast fat-saturated FRFSE image (TR, 2200 ms; TE, 81 ms; flip angle, 90 degrees) reveals no focal liver lesions (dotted circle). (b) Postcontrast fat-saturated FRFSE image using superparamagnetic iron oxide reveals a high signal 12 mm-diameter nodule in segment 1 of the liver (arrow). (c) This nodule shows higher contrast to the surrounding liver parenchyma in the postcontrast fast spoiled gradient-echo image (TR, 190 ms; TE, 10 ms; flip angle, 70 degrees) compared to the FRFSE image (arrow). FRFSE, fast recovery fast spin-echo.

GBCAs are extracellular contrast agents that circulate in the blood and slowly diffuse into the extracellular space. They are eventually excreted by the kidneys, similar to iodinated contrast media.⁹ GBCAs have an advantage over iodinated contrast media in patients with mild to moderate renal dysfunction without dialysis due to their lower risk of contrast-induced nephropathy.¹⁰ A previous meta-analysis revealed that GBCA-enhanced MRI demonstrated higher sensitivity and accuracy than contrast-enhanced CT in detecting hepatocellular carcinoma (HCC).¹¹ Moreover, GBCA-enhanced MRI demonstrated similar sensitivity to contrast-enhanced CT when detecting liver metastases from cancers of the gastrointestinal tract.¹²

SPIO is a liver-specific contrast agent that is taken up by Kupffer cells and was introduced in Japan in 1998. In a systemic review and meta-analysis, SPIO-MRI exhibited high diagnostic ability for differentiating between HCC and benign hepatic lesions with pooled sensitivity and specificity of 0.85 and 0.87, respectively.¹³ In a previous report, SPIO-MRI and contrast-enhanced CT showed similar diagnostic accuracy, sensitivity, and positive predictive value for the detection of HCC in patients with hepatitis B cirrhosis.¹⁴ In another study, SPIO-MRI showed a higher area under the receiver operating characteristic curve (AUROC) (mean, 0.916) compared to that of contrast-enhanced CT (mean, 0.685) in detecting liver metastases from colorectal carcinoma, especially small lesions (≤ 1 cm).¹⁵ However, it is difficult to evaluate blood flow using SPIO-MRI. Currently, the use of SPIO is limited, such as in patients with impaired renal function or allergic to GBCAs or gadoxetic acid.

Gadoxetic acid is a liver-specific contrast agent that is taken up by hepatocytes and was introduced in Japan in 2008. Currently, gadoxetic acid is used almost exclusively for liver MRI examinations in daily clinical practice in Japan because of its high performance in lesion detection and characterization.^4,16–18 Traditionally, sensitivity is important in Japan because the Japan Society of Hepatology (JSH) guideline recommends liver resection or locoregional therapy based on early detection of HCC.¹⁹ However, in the United States and Europe, liver transplantation is an important treatment option; thus, specificity is important. This is also evident in the diagnostic criteria for HCC. Hepatobiliary phase (HBP) hypointensity is treated as equivalent to washout in the JSH guideline¹⁹ (Fig. 3); whereas, the Liver Imaging Reporting and Data System version 2018 (LI-RADS v2018) states that washout should be judged only in the portal venous phase (PVP).²⁰ A core component of surveillance is screening with abdominal ultrasonography and tumor marker tests, which are repeated every 3–6 months. This regular screening procedure may be combined with dynamic CT or gadoxetic-acid-enhanced MRI in extremely high-risk patients including those with cirrhosis. Furthermore, dynamic CT or gadoxetic-acid-enhanced MRI can be performed for cirrhotic patients every 6–12 months based on the current JSH guideline.¹⁹

Fig. 3.

Hepatocellular carcinomas in a 77-year-old man with liver cirrhosis due to hepatitis C virus infection. (a) Precontrast and (b) arterial phase gadoxetic-acid-enhanced MRI reveal hypervascular nodules in segment 6 of the liver (arrows). (c) These nodules show no washout in the portal venous phase (dotted circle), whereas (d) they show hypointensity in the hepatobiliary phase measuring 13 mm in diameter (arrowheads). These nodules are diagnosed as hepatocellular carcinomas according to the Japan Society of Hepatology guideline, whereas they scored LR-4 according to the Liver Imaging Reporting and Data System version 2018.

Gadoxetic-acid-enhanced MRI has a better diagnostic ability than contrast-enhanced CT, particularly for small HCCs²¹ (Fig. 4). Previous meta-analyses revealed that the pooled sensitivity and specificity were 0.85–0.92 and 0.89–0.96, respectively, with AUROC of 0.90–0.99 for all HCCs diagnosed using gadoxetic-acid-enhanced MRI.^21–29 In other studies, gadoxetic-acid-enhanced MRI showed high diagnostic performance for small HCCs (< 2 cm), with sensitivity and AUROC of 0.79–0.94 and 0.97–0.99, respectively.^21–23 Since surgery is indicated for colorectal adenocarcinoma when liver metastases are completely resectable, it is crucial to accurately assess the location and number of metastases before treatment is initiated.

Fig. 4.

Hepatocellular carcinoma in a 62-year-old man with liver cirrhosis due to nonalcoholic steatohepatitis. (a) Arterial and (b) portal venous phase contrast-enhanced CT reveal no focal liver lesions (dotted circles). (c) Arterial and (d) HBP gadoxetic-acid-enhanced MRI performed 3 weeks later reveal a hypervascular nodule that shows hypointensity in HBP measuring 7 mm in diameter in segment 5 of the liver (arrows). Gadoxetic-acid-enhanced MRI has a better diagnostic ability than contrast-enhanced CT. HBP, hepatobiliary phase.

Gadoxetic-acid-enhanced MRI has better diagnostic performance than CT in detecting colorectal liver metastases (Fig. 5).³⁰ According to Sofue et al.,³¹ adding gadoxetic-acid-enhanced MRI to CT changed 33% of the planned operation strategy. Despite the fact that HBP may be useful in detecting small lesions, it is difficult to detect small lesions near blood vessels due to the low signal intensity of intrahepatic vessels. Most institutions in Japan now routinely use diffusion-weighted imaging (DWI) for tumor screening in liver MRI because of its ability to detect small lesions. DWI demonstrates liver metastases as clear high signal intensity with the suppressed intrahepatic vascular signal; thus, combining diffudion-weighting (DW) with HBP images significantly improves the detection of colorectal liver metastases (Fig. 6). Moreover, combined DW and HBP images have the highest sensitivity for detecting liver metastases, especially for small metastases (< 1 cm). The sensitivity for DWI, HBP image, and combined DW and HBP images were 0.87, 0.91, and 0.96, respectively, for all lesion sizes, and 0.69, 0.83, and 0.91, respectively, for small metastases (< 1 cm).³² Previous meta-analysis revealed that the pooled sensitivity and specificity were 0.87–1.00 and 0.80–0.98, respectively, with an AUROC of 0.98 for the diagnosis of colorectal liver metastases using gadoxetic-acid-enhanced MRI with DWI.^32–34

Fig. 5.

Colorectal liver metastases in a 63-year-old man without chronic liver diseases. (a) Contrast-enhanced CT reveals a hypoattenuating 11 mm-diameter nodule in segment 6 of the liver (arrow). This nodule shows hyperintensity in fat-saturated T2-weighted imaging and (c) DWI, whereas it shows hypointensity in (d) HBP gadoxetic-acid-enhanced MRI performed the same day (arrows). Another small 5 mm-diameter nodule is detected via gadoxetic-acid-enhanced MRI in segment 6 of the liver (arrowheads). This small nodule cannot be detected by CT (dotted circle). Gadoxetic-acid-enhanced MRI has better diagnostic performance than CT in detecting colorectal liver metastases. The combination of DWI and HBP gadoxetic-acid-enhanced MRI increases diagnostic accuracy. DWI, diffusion-weighted imaging; HBP, hepatobiliary phase.

Fig. 6.

Liver metastasis from esophagogastric junctional cancer in a 68-year-old man without chronic liver diseases. (a) It is difficult to detect metastases in the HBP gadoxetic-acid-enhanced MRI (dotted arrows). (b) In contrast, DWI clearly shows hyperintense nodules (arrows). Small lesions near blood vessels are difficult to detect because intrahepatic vessels are also depicted with low signal intensity in HBP. DWI, diffusion-weighted imaging; HBP, hepatobiliary phase.

Gadoxetic-acid-enhanced MRI is also useful for detecting liver metastases from pancreatic ductal adenocarcinoma (PDAC). If liver metastases are detected by imaging, surgical resection for PDAC is not indicated. Similar to colorectal liver metastases, metastatic tumors from PDAC usually exhibit early ring-like enhancement and delayed enhancement in their central portions on dynamic CT or MRI. Occasionally, we have encountered unexpected cases of liver metastases that have been misdiagnosed as pseudolesions or microabscesses because they initially appeared as arterioportal shunts. The HBP images clearly show low signal intensity in such a metastasis (Fig. 7). Gadoxetic-acid-enhanced MRI is equivalent to dynamic CT in its ability to depict PDAC and has better sensitivity for the detection of liver metastases, especially small lesions.³⁵

Fig. 7.

Liver metastasis from pancreatic ductal adenocarcinoma in an 83-year-old woman without chronic liver diseases. A small 5mm-diameter nodule is observed in segment 6 of the liver. This nodule cannot be detected in the AP (dotted circle in a) and shows slight hypoattenuation in the PVP (dotted arrow in b). It is difficult to detect via CT. (c) AP gadoxetic-acid-enhanced MRI reveals a wedge shaped hyper-enhanced area that can be misdiagnosed as an arterioportal shunt (arrowhead). (d) HBP imaging clearly shows a hypointense nodule (arrow). Gadoxetic-acid-enhanced MRI has better sensitivity for detecting liver metastases from pancreatic ductal adenocarcinoma. AP, arterial phase; HBP, hepatobiliary phase; PVP, portal venous phase.

Comparing these different types of contrast agents, gadoxetic acid has the highest diagnostic performance for HCC, followed by SPIO and GBCAs.^36,37 However, gadoxetic acid and SPIO are comparable in diagnostic performance for liver metastases, followed by GBCAs.^15,38

Assessment for treatment response of locoregional therapy

Dynamic CT and MRI are recommended follow-up tools for malignant liver lesions with locoregional therapy, including locoablative, transcatheter, and external radiation therapies in both the JSH guideline and LI-RADS v2018. (Fig. 8)^19,20 Furthermore, the JSH guideline recommends follow-up with dynamic CT or MRI every 6–12 months following treatment; in contrast, LI-RADS recommends follow-up posttreatment imaging at 1, 3, 6, 9, and 12 months following locoregional therapy, and approximately every 3–6 months thereafter. Both GBCAs and gadoxetic acid showed similar diagnostic performance in LI-RADS treatment response criteria.³⁹ Previous meta-analyses including contrast-enhanced CT, GBCA-enhanced MRI, and gadoxetic-acid-enhanced MRI revealed that the pooled sensitivity and specificity were 0.56–0.81 and 0.87–0.95, respectively, with an AUROC of 0.80–0.96 for the diagnosis of viable tumor after locoregional therapy including transcatheter arterial chemoembolization (TACE), radiofrequency ablation, percutaneous ethanol injection therapy, and microwave ablation by LI-RADS treatment response criteria with substantial inter-reader reliability.^39–42 In addition, CT and MRI can be comparable in their diagnostic performance for the assessment of local recurrences using LI-RADS treatment response criteria;³⁹ however, it can be difficult to diagnose local recurrence on CT after TACE because of the high concentration of lipiodol accumulated in the lesion. Since lipiodol does not show high signal intensity on MRI, MRI is more comprehensive and accurate in evaluating local recurrence after TACE (Fig. 9).⁴³

Fig. 8.

Recurrent hepatocellular carcinoma after radiofrequency ablation in an 88-year-old man with non-B non-C liver cirrhosis. (a) Precontrast image shows post-ablation lesion in segment 8 of the liver. (b) Arterial phase and (c) portal venous phase contrast-enhanced MRI reveal a 10 mm-diameter nodule that shows arterial phase hyperenhancement and washout at the right periphery in the post-ablation lesion (arrows). This lesion meets the viable criteria of the Liver Imaging Reporting and Data System. (d) This nodule clearly shows hypointensity in the HBP. HBP, hepatobiliary phase.

Fig. 9.

Recurrent hepatocellular carcinoma after transcatheter arterial chemoembolization in a 62-year-old man with liver cirrhosis due to nonalcoholic steatohepatitis. (a) Arterial phase contrast-enhanced CT reveals a focal defect of lipiodol accumulation (arrowhead) 15 months after transcatheter arterial chemoembolization. Nodular enhancement is present (dotted arrow); however, it is difficult to detect it on CT because of the high concentration of lipiodol accumulated in the lesion. (b) This nodular enhancement can be clearly detected in arterial phase contrast-enhanced MRI (arrow) because lipiodol does not show high signal intensity on MRI.

Stereotactic body radiation therapy (SBRT) is one of the treatment options for HCC; however, it is not mentioned in the JSH guideline. Although LI-RADS treatment response criteria can be applied after SBRT, it should be noted that there are posttreatment changes specific to SBRT. Arterial phase hyperenhancement (APHE) with or without washout may persist in the first 6 months following SBRT and gradually decrease over time. Persistent APHE soon after SBRT does not necessarily indicate a residual tumor, and gradually reduces in size over 6–12 months.⁴⁴ The liver parenchyma surrounding the target lesion undergoes temporal imaging changes following SBRT; this is referred to as focal liver reaction (FLR). The acute phase (< 3 months after SBRT) manifests on imaging as APHE conforming to the shape of the non-ablative dose irradiated field with persistent or hypo-enhancement on the PVP. The subacute phase (3–6 months after SBRT) manifests on imaging as hypo-enhancement and hyperenhancement on PVP and delayed phase (DP), respectively. The chronic phase (> 6 months after SBRT) manifests on imaging as minimal APHE with progressive hyperenhancement on DP with atrophy or capsular retraction of the liver due to fibrosis.⁴⁴ At 1 month after SBRT, FLR is depicted as a distinct low signal intensity area on HBP (Fig. 10). The threshold dose of FLR is significantly correlated with baseline liver function. Sanuki et al.⁴⁵ reported median threshold doses of FLR of 30.5 Gy and 25.2 Gy for patients with Child-Pugh score A and B disease, respectively.

Fig. 10.

SBRT for recurrent hepatocellular carcinoma after radiofrequency ablation in an 88-year-old woman with liver cirrhosis due to hepatitis C virus infection. (a) AP and (b) PVP contrast-enhanced MRI reveal a 20 mm-diameter nodule that shows AP hyperenhancement and washout in the periphery of the post-ablation lesion (arrows). (c) This nodule clearly shows hypointensity in the hepatobiliary phase. (d) AP and (e) PVP contrast-enhanced MRI 5 months after SBRT (48 Gy in four fractions) reveal that the nodule decreased in size and disappeared lesional enhancement (arrowheads). This lesion meets the nonviable criteria of the Liver Imaging Reporting and Data System. The liver parenchyma surrounding the target lesion shows AP hyperenhancement and persistent enhancement on the PVP (dotted arrows). (f) This area is depicted as a distinct low signal intensity area on the hepatobiliary phase (dotted arrows). AP, arterial phase; PVP, portal venous phase; SBRT, stereotactic body radiation therapy.

Potential of gadoxetic acid as an imaging biomarker

HBP hypointense nodules without APHE constitute an entity that is unique in gadoxetic-acid-enhanced MRI.⁴⁶ They are considered to be borderline nodules, with this category including dysplastic nodules and early HCCs that are at risk of transformation into progressed HCC. The previous meta-analysis revealed that the pooled overall rate of hypervascularization of HBP hypointense nodules without APHE was 28.2%. In addition, the pooled 1-, 2-, and 3-year cumulative incidence rates were 18.3%, 25.2%, and 30.3%, respectively.⁴⁷ Moreover, baseline size greater than 9–10 mm, T2 hyperintensity, restricted diffusion, and previous history of HCC were reported as risk factors for hypervascularization.^47,48 In addition to multistep carcinogenesis, which was previously thought to be the predominant cause of HCC development, the “imaging de novo” or “imaging-occult” processes are now considered to be the cause present in a larger proportion than previously thought.^49,50 Patients with HBP hypointense nodules without APHE have a high risk of hypervascular HCC development not only from those nodules but also from any part of the liver.^50–52 Based on these results, HBP hypointense nodules without APHE are an imaging biomarker for patients at high risk of developing hypervascular HCCs.

HBP hypointense nodules without APHE in pretreatment gadoxetic-acid-enhanced MRI should also be recognized as a significant risk factor for increased recurrence after curative treatment for HCCs. The previous meta-analysis revealed that the overall pooled hazard ratio for intrahepatic distant recurrence after ablation or hepatectomy was 2.44 in patients with HBP hypointense nodules without APHE.⁵³ HBP hypointense nodules without APHE can be an indicator of both early and late recurrence after surgery for HCCs.^54,55 Considering all these observations, HBP hypointense nodules without APHE may be a sign of accelerated hepatocarcinogenesis in the entire liver, rather than a simple precursor of progressed HCC. Clinically, it is important to determine the presence of HBP hypointense nodules without APHE for decision-making in patients with CLD.

Concerns about gadoxetic acid

Insufficient quality on the arterial phase

Gadoxetic-acid-enhanced MRI is a relatively useful tool for the diagnosis of focal liver lesions; however, it has some disadvantages. Hepatic dynamic MRI is typically performed using a 3D gradient-echo sequence with > 15s of breath-holding. However, breath-holding for > 10s is challenging for some patients. The image quality at the arterial phase (AP) of gadoxetic-acid-enhanced MRI may be insufficient due to transient severe motion artifacts. Several studies have reported that transient respiratory motion artifacts or truncation artifacts in the AP can be observed more frequently in dynamic MRI with gadoxetic acid than that with GBCAs,^56–58 resulting in nondiagnostic image quality in the AP.

The scan timing of the AP is another concern in dynamic MRI with gadoxetic acid. The injection volume of gadoxetic acid is half that of GBCAs, which results in a shorter duration when using the same injection rate. Multiphasic AP imaging can be used to treat these problems, as it can provide at least one image set with reduced or no artifacts and appropriate scan timing of the AP.⁵⁹ Another strategy involves free-breathing dynamic MRI that can reduce the risk of respiratory artifacts. Stack-of-stars acquisition that applies Cartesian sampling along the z-axis and radial sampling along the xy-plane is used for free-breathing dynamic MRI. Inappropriate scan timing is also not an issue in free-breathing dynamic MRI with stack-of-stars acquisition because it is possible to obtain continuous images from the AP to the PVP and the transitional phase (TP) with one scan.^60,61 However, it is difficult for free-breathing dynamic MRI with stack-of-stars acquisition to outperform breath-holding Cartesian sampling and multiphasic AP imaging in terms of image quality in patients who can hold their breath for a prolonged period. Therefore, Cartesian sampling or multiphasic AP imaging may be better than stack-of-stars acquisition as a routine sequence. As a result, free-breathing hepatic dynamic MRI with stack-of-stars acquisition should be used in patients with a history of inappropriate AP scan timing or poor breath-hold with Cartesian sampling and multiphasic AP imaging.

Insufficient uptake of gadoxetic acid on the HBP

The liver is not well enhanced in the HBP in patients with severe liver dysfunction because of decreased uptake of gadoxetic acid due to liver dysfunction (Fig. 11).^62,63 In such cases, contrast-enhanced CT or MRI with GBCAs may be more useful for the diagnosis of focal liver lesions. Usually in the HBP, the liver parenchyma shows marked higher signal intensity than the portal vein or spleen. In contrast, in patients with severe liver dysfunction, the signal intensity of the liver parenchyma is not sufficiently elevated, and the contrast with the portal vein and spleen is reduced.

Fig. 11.

Hepatocellular carcinomas in a 59-year-old man with liver cirrhosis due to hepatitis C virus infection. (a) Arterial phase and (b) portal venous phase contrast-enhanced (CT reveals a 15 mm-diameter hypervascular nodule that shows a washout appearance in segment 3 of the liver (arrows). (c) Arterial phase gadoxetic-acid-enhanced MRI also reveals the hypervascular nodule; however, (d) this nodule shows iso intensity compared to the surrounding liver parenchyma in the HBP (dotted circle). In cases with good uptake of gadoxetic acid in the HBP, the liver parenchyma shows a distinctly higher signal than the spleen and intrahepatic vessels (see Fig. 3d), whereas in cases with severely impaired liver function, the liver parenchyma shows a signal comparable to that of the spleen and intrahepatic vessels because of decreased gadoxetic acid uptake. HBP, hepatobiliary phase.

Obscuration of enhancing capsule

An enhancing capsule is one of the major features of HCC.²⁰ It is observed as a ring-like enhancement in the PVP on gadoxetic-acid-enhanced MRI; however, it is often difficult to detect an enhancing capsule at the PVP because of the uptake of gadoxetic acid into hepatocytes even in the PVP.⁶⁴ Previously, studies have reported that the frequency of enhancing capsules in the PVP acquired using GBCAs and gadoxetic acid was 23%–64%^65–68 and 17%–49%,^68–70 respectively. Enhancing capsules of HCCs can be detected more frequently by the combination of multiphasic AP and PVP imagings, which enables a confident diagnosis of HCCs in accordance with the LI-RADS system.⁷¹

Pseudo-washout in the TP

Pseudo-washout of hemangioma is another area of concern. Prolonged enhancement in the DP is the key imaging finding of hepatic hemangioma. The lack of prolonged enhancement is called pseudo-washout due to the uptake of gadoxetic acid in the surrounding normal liver parenchyma in the TP (Fig. 12).⁷² Small hemangiomas tend to decrease in signal intensity in the PVP and show pseudo-washout in the TP.^72–74 Such hemangiomas can mimic small HCCs; however, T2-weighted imaging (T2WI) and DWI may provide additional information to enhance the confidence to exclude small HCCs.⁷⁴

Fig. 12.

Hepatic hemangiomas in a 38-year-old man without chronic liver diseases. (a) Precontrast and (b) arterial phase gadoxetic-acid-enhanced MRI reveal peripherally enhancing nodules in segments 8 (45 mm in diameter) and 7 (20 mm in diameter) of the liver (arrow and arrowhead). (c) These nodules show centripetal enhancement in the portal venous phase. (d) These findings are compatible with a cavernous hemangioma; however, these nodules show hypointensity compared to the surrounding liver parenchyma (pseudo-washout) in the transitional phase.

Gadolinium deposition

Since Kanda et al.⁷⁵ reported an association between hyperintensity in the dentate nucleus and globus pallidus on T1WI and a history of GBCAs administration in 2014, many findings about gadolinium deposition in the brain have been accumulated.⁷⁵ Even in patients with normal renal function, increased signal intensity in the dentate nucleus and globus pallidus on unenhanced T1WI has been reported to be positively correlated with previous exposure to linear chelate GBCAs but not to macrocyclic chelate GBCAs.⁷⁶ Thus, gadolinium from linear chelate GBCAs is more readily retained in the body than that from the macrocyclic chelate GBCAs. Aside from gadoxetic acid, linear chelate GBCAs are rarely used in Japan. Although gadoxetic acid is a linear chelate contrast agent, it exhibits greater thermodynamic stability than some of the other linear chelate GBCAs.^77,78 Moreover, gadoxetic acid contains half the concentration of gadolinium (0.25 mmol/mL) as GBCAs (0.50 mmol/mL), and its dosage of administration (0.1 mL/kg) is half that of other GBCAs (0.2 mL/kg). Thus, investigations with gadoxetic acid utilize only a quarter the amount of gadolinium as those with other GCBAs (GBCAs, 0.1 mmol/kg vs. gadoxetic acid, 0.025 mmol/kg). Therefore, gadoxetic acid is less likely to be deposited in the brain than other linear GBCAs.⁷⁹

Long scan time and high cost

Long scan time (30–40 minutes) and cost for gadoxetic-acid-enhanced MRI are major problems limiting its use as a screening tool. Abbreviated MRI is a technique first reported in 2014 for breast MRI,⁸⁰ which reduced scan time by taking only the minimum sequence of imaging necessary for the examination. This technique can be performed in less than 10 minutes from the patient entering the examination room to leaving the room. Different abbreviated MRI approaches have been proposed, including non-contrast abbreviated MRI protocols (T2WI and DWI), dynamic abbreviated MRI protocols (dynamic T1WI using GBCAs with or without T2WI), and gadoxetic acid abbreviated MRI (HBP, T2WI, and DWI).⁸¹ DWI and HBP gadoxetic-acid-enhanced MRI are the most useful sequences for detecting focal liver lesions; thus, gadoxetic acid abbreviated MRI may be the best abbreviated MRI protocol for liver MRI. In gadoxetic acid abbreviated MRI, gadoxetic acid is injected outside the MRI room without dynamic study, which reduces complexity and potentially improves efficiency.

Most published abbreviated MRI studies have retrospectively evaluated the performance of simulated abbreviated MRI examinations that picked up T2WI, DWI, and HBP gadoxetic-acid-enhanced MRI from full MRI examinations. The previous meta-analysis revealed that the pooled sensitivity and specificity were 0.86–0.87 and 0.94–0.96, respectively, with an AUROC of 0.95 for the detection of HCC using gadoxetic acid abbreviated MRI.^82,83 Another meta-analysis of gadoxetic acid abbreviated MRI reported an excellent inter-reader agreement (κ = 0.98) for the detection of HCC.⁸⁴

Few papers have examined the usefulness of abbreviated MRI for searching liver metastases. The reported sensitivity of gadoxetic acid abbreviated MRI was 0.93–0.94 with an AUROC of 0.91–0.95 for the detection of colorectal liver metastases.⁸⁵ Gadoxetic acid abbreviated MRI may also be useful in the detection of liver metastases from PDAC;⁸⁶ however, its clinical use should be carefully considered since MR cholangiopancreatography (MRCP), one of the most important sequences for the diagnosis of PDAC, cannot be taken before the injection of gadoxetic acid. MRCP should be obtained before the injection of gadoxetic acid because gadoxetic acid decreases the signal intensity of the biliary trees on MRCP.⁸⁷

Reduced cost is another advantage of abbreviated MRI. Canellas et al. estimated that the cost of gadoxetic acid abbreviated MRI represented 59% of the full MRI examinations.⁸⁵ Therefore, if its clinical usefulness is fully explored, gadoxetic acid abbreviated MRI could become widely employed for the screening of HCC or liver metastases.

Clinical Needs for Diffuse Liver Diseases

Assessment of hepatic steatosis and iron overload

Dixon technique

Globally, NAFLD affects up to one-fourth of the population and increases hepatic and cardiometabolic risks with consequent adverse outcomes. NAFLD can be categorized histologically into the nonalcoholic fatty liver (NAFL) or NASH. NAFL is defined as the presence of ≥ 5% hepatic steatosis without evidence of hepatocellular injury. In contrast, NASH is defined as the presence of ≥ 5% hepatic steatosis and liver inflammation with hepatocellular injury with or without fibrosis.⁸⁸ This may lead to advanced fibrosis that is associated with an extremely high risk of liver-related morbidity and mortality.

A liver biopsy is essential for the diagnosis of NASH because clinical, biochemical, or imaging findings cannot differentiate NASH from NAFL; in contrast, MRI is used for screening NAFLD.⁸⁸ Proton density fat fraction (PDFF) assessed with the Dixon technique is useful for quantifying hepatic steatosis. Iron is a confounding factor in measuring the fat fraction of the liver. Iterative decomposition of water and fat with echo asymmetry and least squares estimation (IDEAL)-IQ can be applied for the correct calculation of PDFF even under T2* decay, where the parametric maps of PDFF (Fig. 13) and R2* are automatically calculated. IDEAL-IQ uses T2* signal decay with a detailed fat peak model in the reconstruction algorithm.⁸⁹ PDFF showed a good correlation with MR spectroscopy which has been accepted as the most accurate noninvasive technique for hepatic steatosis.^90,91 Recent meta-analyses have reported high diagnostic performance for the classification of histological steatosis in patients with NAFLD. The pooled sensitivity and specificity were 0.74–0.83 and 0.89–0.90, respectively, with an AUROC of 0.90–0.91 for the staging of moderate steatosis (S ≥ 2) and 0.74–0.79 and 0.87–0.89, respectively, with an AUROC of 0.90–0.92 for the staging of severe steatosis (S3) in PDFF.^91–93 PDFF has a higher diagnostic ability for hepatic steatosis as compared to transient elastography (TE)-controlled attenuation parameter (CAP).^93,94 CAP value is based on the properties of ultrasonic signals acquired using TE.⁹⁴ In addition, CAP increases with the fat content and can detect > 5% of hepatic steatosis.⁹⁵

Fig. 13.

Examples of PDFF of a 52-year-old woman with chronic hepatitis B without hepatic steatosis (upper row), a 51-year-old woman with NALFD with moderate hepatic steatosis (middle row), and a 56-year-old woman with NAFLD with severe hepatic steatosis (lower row). No signal loss in the liver parenchyma is observed between (a) in-phase and (b) in opposed-phase images. (c) PDFF shows hypointensity throughout the liver. Fat fraction is calculated as 2.2% in this patient. Mild signal loss is observed on (e) opposed-phase image compared to (d) in-phase image. (f) PDFF shows higher signal intensity compared to (c) throughout the liver. Fat fraction is calculated as 11.1% in this patient. Evident signal loss is observed on (h) opposed-phase image compared to (g) in-phase image. PDFF shows higher signal intensity compared to (c) and (f) throughout the liver. Fat fraction is calculated as 46.3% in this patient. NAFLD, nonalcoholic fatty liver disease; PDFF, proton density fat fraction.

Hepatic iron content can also be evaluated by R2* maps obtained simultaneously with IDEAL-IQ (Fig. 14). Moreover, hepatic iron overload causes chronic hepatocellular injury.⁹⁶ Ferritin is a storage protein for iron. In patients with NAFLD, hyperferritinemia can be related to inflammation; therefore, hepatic iron overload should be evaluated noninvasively.⁹⁷

Fig. 14.

Examples of an R2* map of a 73-year-old man with liver cirrhosis due to hepatitis C virus infection without hepatic iron overload (upper row) and an 82-year-old woman with hemochromatosis with severe hepatic iron overload (lower row). No signal loss in the liver parenchyma is observed between (a) in-phase and (b) in opposed-phase images. (c) T2*-weighted image shows no signal decrease in the liver parenchyma. (d) R2* map shows hypointensity throughout the liver. R2* value is calculated as 53.3s^–1 in this patient. (e) In-phase image shows a diffuse signal loss in the liver parenchyma compared to the (f) opposed-phase image due to severe iron deposition. (g) Liver parenchyma shows lower signal intensity in the T2*-weighted image compared to (c). (h) R2* map shows evident hyperintensity throughout the liver. R2* value is calculated as 928.0s^–1 in this patient.

MR spectroscopy (MRS)

MRS is a noninvasive method of accurately quantifying intrahepatic lipid levels and can detect even small amounts of intrahepatic lipid accumulation.⁹⁸ Similar to conventional MR imaging, MRS applies radiofrequency pulses, measures the signal from the tissue, and performs Fourier transform. However, although MR imaging uses frequency variation to spatially localize the signal to voxels and create cross-sectional images, MRS uses frequency (position on the x-axis) to isolate and characterize the actual metabolites and chemicals in the voxel. In addition, the signal intensity (position on the y-axis) and line width provide chemical information about the tissue, allowing the relative amounts of chemicals to be determined. Therefore, MRS applies the law of chemical shifts to obtain qualitative and quantitative information about the chemicals present between water and fat. In MRS, the parts per million (ppm) scale represents the resonance (position on the x-axis) of a chemical. Water is located at 4.26 ppm, which is commonly used as the reference frequency for in vivo abdominal MRS; fat is at 0.9–1.4 ppm.⁹⁹ Previous meta-analysis revealed that MR imaging and MRS showed high pooled sensitivity and specificity of 0.89 and 0.84, respectively, for the detection of hepatic steatosis.¹⁰⁰ Since MRS is limited to evaluating only small areas of the liver and complicated analysis methods, PDFF, which can be performed easily, has become widely accepted for the evaluation of hepatic steatosis.¹⁰¹

Assessment of hepatic fibrosis

T1WI and T1 mapping pre and post-gadoxetic acid administration

Gadoxetic acid is taken up by hepatocytes and has the effect of shortening the T1 value. However, the linearity between the signal intensity in MR images and the concentration of gadoxetic acid is not high.¹⁰² On the other hand, the concentration of gadoxetic acid and R1 (1/T1) show high linearity.¹⁰² Therefore, measuring the amount of change in the T1 value can accurately represent the amount of change in the concentration of gadoxetic acid. Thus, T1 relaxation time measurement on gadoxetic-acid-enhanced MRI may be a potential biomarker for liver fibrosis.¹⁰³ T1 mapping is a technique that measures tissue-specific T1 values and is useful for quantitatively evaluating the degree of contrast effect. In addition, it can assess hepatic fibrosis and inflammation without contrast agents. Both fibrosis and inflammation lead to an increase in liver extracellular fluid, the amount of which can be measured by an increase in T1 relaxation time.⁹⁷ Iron overload can be quantified by acquiring T2* in parallel in the same slice since iron overload can be a confounding factor by decreasing T1 relaxation time measurements. Iron-corrected T1 mapping is already used in clinical trials as an endpoint for evaluating diffuse liver disease and monitoring treatment response.^104–106 T1 mapping has potential applications in the evaluation of hepatocyte function because it permits quantitative evaluation of the kinetics of gadoxetic acid uptake and excretion into the biliary system.⁴

DWI

DWI is a technique for imaging molecular movement or diffusivity and can measure the apparent diffusion coefficient (ADC) value that represents the diffusivity of the subjects. Since the liver is a highly vascularized organ, perfusion-related effects are not negligible for diffusivity assessment by DWI.¹⁰⁷ Intravoxel incoherent motion (IVIM) is an imaging technique used for the simultaneous estimation of perfusion-related and pure molecular diffusivities by analyzing the signal decay of multi-b-value DWIs.¹⁰⁸ Three parameters—pure molecular diffusivity (D), perfusion-related diffusivity (D*), and perfusion-related diffusion fraction (f)—can be obtained by the IVIM model. Fibrosis and cirrhosis can significantly change the histopathological structures of the liver; therefore, ADC or IVIM parameters can be largely altered. Although DWI and IVIM have been reported to be useful in staging liver fibrosis, their diagnostic ability is not better than that of MR elastography (MRE).^109,110 The pooled sensitivity and specificity were 0.63–0.82 and 0.78–0.86, respectively, with an AUROC of 0.79–0.88 for the staging of significant fibrosis (F ≥ 2), 0.71–0.88 and 0.68–0.84, respectively, with an AUROC of 0.81–0.89 for the staging of severe fibrosis (F ≥ 3), and 0.80–0.96 and 0.69–0.77, respectively, with an AUROC of 0.80–0.90 for the staging of cirrhosis (F4) in DWI and IVIM.^{109,111–114}

Elastography

Cirrhosis is the most important risk factor for HCC; therefore, it is important to accurately assess advanced fibrosis. For many patients, fibrosis stages are monitored using noninvasive tests, including elastography. Although ultrasound elastography is more frequently used in clinical practice, MRE is also gaining traction. In addition, MRE creates a stiffness map (elastogram) using low-frequency vibrations (Fig. 15). A vibrator placed outside the examination room produces a pneumatic vibration that is delivered to the passive driver using a plastic cylinder. The driver then transfers the vibration to the liver via the chest wall. The MR scanners automatically generate liver stiffness maps by processing the acquired propagating shear wave images according to a 2D or 3D inversion algorithm.

Fig. 15.

Examples of MR elastography of a 54-year-old woman with nonalcoholic fatty liver disease (upper row) and a 52-year-old man with alcoholic liver cirrhosis (lower row). (a) Fat-saturated T1WI shows no morphological change. (b) Wave image shows propagating shear wave in the liver. (c) Elastograms are automatically generated by processing the acquired wave images according to a 2D or 3D inversion algorithm. The mean liver stiffness value is calculated as 2.2 kPa in this patient. (d) Cirrhotic change is observed in fat-saturated T1WI. (e) Wave image shows a larger amplitude of the waves propagating through the liver than that in Fig. 8b. (f) Elastogram shows that the mean liver stiffness value of the cirrhotic liver is significantly higher than that of the non-cirrhotic liver. The mean liver stiffness value is calculated as 11.6 kPa in this patient. T1WI, T1-weighted image.

Recent studies have reported that MRE is a highly reproducible and useful tool for the noninvasive staging of liver fibrosis. Although more reports indicated that MRE had a higher diagnostic performance for liver fibrosis than ultrasound elastography,^115–118 some reports showed no significant difference in diagnostic performance between the two.^119,120 Recent meta-analyses of MRE have reported excellent diagnostic performance for the staging of liver fibrosis. The pooled sensitivity and specificity were 0.78–0.91 and 0.85–0.98, respectively, with an AUROC of 0.91–0.97 for the staging of significant fibrosis (F ≥ 2), 0.83–0.95 and 0.83–0.94, respectively, with an AUROC of 0.92–0.97 for the staging of severe fibrosis (F ≥ 3), and 0.80–1.00 and 0.75–0.93, respectively, with an AUROC of 0.90–0.99 for the staging of cirrhosis (F4).^{109,111,121–125} For clinical purposes, following cut-off values are recommended: 3.0 kPa for F1, 3.5 kPa for F2, 4.0 kPa for F3, and 4.5 kPa for F4.¹²⁶

In contrast, most imaging studies of liver stiffness have been performed with ultrasound elastography. It can quantitatively assess liver stiffness using an external force, such as a mechanically induced impulse in TE or ultrasound-induced focused radiation force impulse in point shear wave elastography (SWE).¹²⁷ Recent meta-analyses have reported the high diagnostic performance of TE and SWE for the staging of liver fibrosis. In TE, the pooled sensitivity and specificity were 0.76–0.82 and 0.65–0.91, respectively, with an AUROC of 0.81–0.91 for the staging of significant fibrosis (F ≥ 2), 0.75–1.00 and 0.74–0.91, respectively, with an AUROC of 0.84–0.91 for the staging of severe fibrosis (F ≥ 3), and 0.69–0.92 and 0.81–0.88, respectively, with an AUROC of 0.84–0.95 for the staging of cirrhosis (F4).^{109,124,125,128–132} In SWE, the pooled sensitivity and specificity were 0.71–0.88 and 0.67–0.83, respectively, with an AUROC of 0.79–0.92 for the staging of significant fibrosis (F ≥ 2), 0.72–0.95 and 0.72–0.87, respectively, with an AUROC of 0.72–0.95 for the staging of severe fibrosis (F ≥ 3), and 0.75–0.89 and 0.84–0.93, respectively, with an AUROC of 0.88–0.94 for the staging of cirrhosis (F4).^{123,124,132–137} Therefore, the diagnostic ability of TE and SWE for staging liver fibrosis cannot be as high as that of MRE.

High reproducibility is another advantage of MRE;^138–140 therefore, MRE is the most robust imaging technique for the assessment of liver fibrosis. However, there are some confounders as well, specifically hepatitis activity and iron deposition. Furthermore, hepatocyte swelling, interstitial edema, increased local blood supply, and inflammatory cell infiltration may increase liver stiffness in patients with acute hepatitis.^141,142 Iron deposition in the liver decreases the signal intensities due to the susceptibility effect. Gradient-echo MRE can be more vulnerable to susceptibility from iron deposition than spin-echo echo-planar imaging (SE-EPI) MRE¹³⁹ (Fig. 16).

Fig. 16.

Examples of MRE of a 65-year-old man with hepatic iron overload. (a) In-phase image shows a higher number of spots with signal loss in the liver parenchyma compared to (b) opposed-phase due to iron deposition. (c) There is no measurable area of liver stiffness value on gradient-echo MRE. Cross-hatched regions on the elastogram are areas of low-confidence data excluded by the processing algorithm. (d) Spin-echo echo-planar imaging MRE produces an elastogram with good image quality. MRE, MR elastography.

Free-breathing and SE-EPI sequences have been developed for MRE. This allows for faster image acquisition and reduces the effects of respiratory motion. The 3D-MRE also provides additional parameters such as volumetric strain, loss modulus, and attenuation.¹²⁶ These parameters are useful for a more detailed assessment of the pathology; however, these parameters are currently under evaluation and await routine clinical application.

T1ρ

T1ρ value has been reported to be a potential MR biomarker for liver function. Chen et al.¹⁴³ reported that the T1ρ values of the liver parenchyma were significantly correlated with albumin, prothrombin time, and Child-Pugh classification. Suyama et al.¹⁴⁴ reported that the T1ρ values of the right liver lobe positively correlated with the SWE values. Takayama et al.¹⁴⁵ reported that T1rho values of the liver parenchyma significantly increased as the severity of liver fibrosis and necro-inflammation progressed. The principle of the T1ρ value is that once the spin magnetization is tilted transversely and then a pulse is applied, the magnetization becomes spin locked and rotates at the frequency of the pulse. The monoexponential decay of magnetization is sampled by imaging at different spin-lock times. The relaxation constant associated with this decay is denoted T1ρ.¹⁴⁶ T1ρ imaging does not require additional equipment or advanced analysis software; thus, it may become a popular tool for the evaluation of hepatic fibrosis in the future.

Conclusion

Liver MRI is an essential modality in clinical practice. It is a comprehensive diagnostic method that can evaluate focal liver lesions and CLDs in a single examination. However, MRI also has some disadvantages, including longer scan times and higher costs. Since MRI cannot be performed in all patients, it is important to combine it with other modalities to manage the treatment plan for liver diseases.