Advances in Functional and Molecular MRI Technologies in Chronic Liver Diseases

Summary

MRI has emerged as the most comprehensive noninvasive diagnostic tool for liver diseases. In recent years, the value of MRI in the liver has been significantly enhanced by a wide range of contrast agents, both clinically available and under development, that add functional information to the anatomically detailed morphological images, or increase the distinction between normal and pathological tissues by targeting molecular and cellular events. Several classes of contrast agents are available for contrast-enhanced hepatic MRI, including 1) conventional nonspecific extracellular fluid (ECF) contrast agents for assessing tissue perfusion; 2) hepatobiliary-specific contrast agents that are taken up by functioning hepatocytes and excreted through the biliary system for evaluating hepatobiliary function; 3) Superparamagnetic iron oxide particles that accumulate in Kuppfer cells; and 4) novel molecular contrast agents that are biochemically targeted to specific molecular/cellular processes for staging liver diseases or detecting treatment responses. The use of different functional and molecular MRI methods allows noninvasive assessment of disease burden, progression, and treatment response in a variety of liver diseases. The multiparametric capability of MR imaging provides the opportunity for high diagnostic performance by combining imaging biomarkers.

Introduction

Chronic liver disease (CLD) is a major public health problem worldwide. The major causes of CLD include hepatic viral infections, chronic exposure to toxins or drugs (e.g., alcohol abuse), chronic alteration of metabolism, and persistent autoimmune reaction. CLD may induce steatosis, iron dysregulation, inflammatory response and/or fibrogenesis. Liver fibrosis, a common feature of almost all causes of progressive CLD, involves the accumulation of collagen, proteoglycans, and other macromolecules within the extracellular matrix. Without removal of exposure to the specific etiology, fibrosis tends to progress, leading to cirrhosis, hepatic dysfunction, and portal hypertension. Therefore, it is crucial to be able to differentiate liver fibrosis stages since treatment decision and monitoring algorithms are related to the degree of fibrosis [1-3].

Liver biopsy is the gold standard for diagnosing and staging CLDs, but is invasive with risk of complications, suffers from sampling errors due to heterogenous distribution of pathology, and inter-observer variation [4, 5]. These drawbacks limit the use of liver biopsy for serial monitoring of disease progression, therapy response assessment, and longitudinal follow-up studies. Blood markers of liver diseases, as a noninvasive alternative to liver biopsy, can be easily obtained, are less affected by sampling error, and can be repeated serially. However, most of these markers lack sensitivity and/or specificity, may detect false negative in end-stage CLDs, and can be confounded by a wide variety of extrahepatic disorders [6, 7]. Thus, development of a noninvasive, accurate, and reproducible test for diagnosis and monitoring of liver diseases and therapy response would be of great value.

Ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI) are routinely used in clinics to detect, stage, and assess treatment response of liver diseases [8]. In recent years, liver MRI has seen tremendous growth. Manifold technical improvements, including the development of higher clinical field strengths, parallel imaging techniques, stronger gradients, and sensitive surface coils, have substantially improved the image quality and speed of liver MRI examinations [9]. MRI is also the most versatile noninvasive modality allowing visualization of a variety of anatomical and functional features [10]. Chemical shift-based water-fat separation methods allow for accurate hepatic fat quantification [11, 12]. T2* weighted images are sensitive to iron accumulation in the liver and the T2* value has been applied to quantify liver iron content in patients with iron loading disorders [13, 14]. Native T1 value (without contrast agents) [15, 16], T1rho relaxation time [17-19], diffusion-weighted imaging [20-22], and MR elastography (MRE) [23-25] have been applied to assess liver fibrosis. Besides these endogenous contrast mechanisms, a wide range of contrast agents, both clinically available and under development, add functional information to the anatomically detailed morphological images, or increase the distinction between normal and pathological tissues by targeting molecular and cellular events. The diagnostic value of contrast-enhanced MRI in liver lesion detection and characterization has been extensively reviewed [26-29] and will not be discussed here. In this review, we will provide an overview of available MRI contrast agents for functional and molecular liver MRI in the context of diffuse liver diseases; discuss their use in clinical and preclinical studies of CLDs to assess disease burden, progression, and treatment response; and point out future directions that will pave the way for their clinical translation or use.

Classification of MRI contrast agents for liver imaging

Contrast agents used in hepatobiliary imaging can be classified into four major categories depending on their biodistribution mechanism: 1) conventional nonspecific extracellular fluid (ECF) contrast agents that accumulate in the vascular space and extracellular extravascular space by a passive distribution mechanism; 2) hepatobiliary contrast agents that are taken up specifically by hepatocytes and partially excreted through the biliary system; 3) superparamagnetic iron oxide (SPIO) particles that are taken up by the reticuloendothelial system (RES), in particular entrapped by resident hepatic macrophages, Kupffer cells; 4) molecular contrast agents that are biochemically targeted to specific molecular processes. Table 1 lists examples of contrast agents from each category. The majority of these agents are gadolinium (Gd) chelates [29-31]. Gd³⁺ is highly paramagnetic resulting in a shortening of the T1 and T2 relaxation times of nearby water protons. The T1 shortening effect of gadolinium chelates predominates under most conditions, which increases the signal intensity of T1-weighted MRI scan in regions where the agents distribute [27, 31]. Therefore, the Gd-based contrast agents are also called positive contrast agents. SPIO particles, on the other hand, are referred as negative contrast agents due to their susceptibility effect which causes local magnetic field inhomogeneities and a resultant signal reduction on T2- or T2*-weighted images [32, 33].

Table 1.

Functional and molecular MRI contrast agents for imaging liver diseases.

AGENT (TRADE NAME)	DISTRIBUTION TYPE/TARGET	DEVELOPMENT STAGE	INDICATION
Gd-DTPA (Magnevist)	ECF	Approved (1988); suspended (EU 2017)	Liver lesions
Gd-DOTA (Dotarem, Clariscan)	ECF	Approved (EU 1989; US 2013)	Liver lesions
Gd-HPDO3A (ProHance)	ECF	Approved (1992)	Liver lesions
Gd-DO3A-butrol (EU: Gadovist; US: Gadavist)	ECF	Approved (EU 1998; US 2011)	Liver lesions
Gd-BOPTA (Multihance)	ECF/heptobiliary	Approved (EU 2005; US 2008)	Liver lesions; hepatocyte function
Gd-EOB-DTPA (EU: Primovist; US: Eovist)	Heptobiliary	Approved (EU 1998; US 2004)	Liver lesions; hepatocyte function; liver fibrosis/cirrhosis (animal studies)
EP-3533	Type I collagen	Animal studies (CCl4, BDL, DEN, CDAHFD); treatment response (rapamycin, farnesoid X receptor agonist, PPAR-α/δ agonist)	Liver fibrosis
CM-101	Type I collagen	Animal studies (BDL, CCl4)	Liver fibrosis
ProCA32.collagen1	Type I collagen	Animal studies (TAA/alcohol, DEN, NASH diet)	Liver fibrosis
Gd-ESMA	Elastin	Animal studies (CCl4)	Liver fibrosis
Gd-Hyd	Allysine/oxidized collagen	Animal studies (CCl4, CDAHFD); treatment response (farnesoid X receptor agonist, PPAR-α/δ agonist)	Liver fibrogenesis, NASH
c(RGDyC)-USPIO	Integrin αvβ3 (HSCs)	Animal studies (CCl4)	Liver fibrogenesis
Gd-P	Fibrin-fibronectin	Animal studies (CCl4)	Liver fibrosis
EP-2104R	Fibrin	Animal studies (DEN)	Liver inflammation
MPO-Gd	Myeloperoxidase	Animal studies (MCD diet); human biopsy samples	Liver inflammation, NASH
SPIO (Endorem, Feridex; Resovist)	RES/Kupffer cells	Approved but withdrawn or discontinued	Liver lesions, fibrosis
USPIO ferumoxytol (Feraheme)	RES/Kupffer cells	Approved (EU 2013; US 2009) only for use in iron replacement therapy	Liver lesions

ECF = extracellular fluid; EU = European Union; US = United States; CCl₄ = carbon tetrachloride; BDL = bile duct ligation; DEN = diethylnitrosamine; TAA = thioacetamide; CDAHFD = choline-deficient L-amino acid-defined high-fat diet; MCD = methionine and choline–deficient; HSC = hepatic stellate cell; NASH = nonalcoholic steatohepatitis; RES = reticuloendothelial system.

Extracellular fluid (ECF) agents

ECF agents are small, hydrophilic chelates of Gd³⁺ that exhibit no protein binding and are used to measure the perfusion, blood flow, and vascularity of liver tissue. ECF agents enter the liver via the hepatic artery and portal vein, rapidly distribute throughout the extracellular space, and are excreted almost entirely by glomerular filtration [27, 30, 34]. Compared to iodinated CT contrast agents, Gd-based ECF agents are detected with orders of magnitude better sensitivity [35], which allows better enhancement of the blood pool at equilibrium phase and better delineation of subtle areas of contrast agent accumulation [36, 37]. ECF agents are the least expensive and the most commonly used contrast agents, and are generally considered safe when administered at low dosage [38, 39]. The ECF agents can be categorized into linear or macrocyclic by their chemical structure. Macrocyclic agents are considered safer than linear agents with respect to any release of Gd ion into the body, and are now more widely used.

Hepatobiliary agents

Hepatobiliary agents are paramagnetic compounds taken up by functioning hepatocytes and excreted in bile. Hepatobiliary agents have been available for over two decades starting from the now discontinued mangafodipir trisodium (Mn-DPDP). Mn-DPDP is the only manganese-based contrast agent ever approved for clinical use, but was withdrawn from all markets due to lack of commercial success. It was administered by slow infusion and provided strong positive liver signal, but was not suitable for dynamic imaging. The two currently available hepatobiliary agents, gadoxetic acid (Gd-EOB-DTPA) and gadobenate dimeglumine (Gd-BOPTA), are both gadolinium based and exhibit higher T1 relaxivity (the ability to change T1 relaxation) than Mn-DPDP or Gd-based ECF agents. Gd-EOB-DTPA is an amphiphilic gadolinium chelate first described in 1991 [40]. Unlike ECF agents, it displays liver-specific properties based on organic anion-transporting polypeptides (OATP1B1 and B3)-mediated hepatic uptake and multidrug resistance protein (MRP2)-mediated biliary excretion by hepatocytes [41-44] (Figure 1A). In healthy subjects, the elimination is 50% hepatic and 50% by glomerular filtration [45, 46]. Under pathological conditions, there are two main factors that determine the cellular uptake and excretion of Gd-EOB-DTPA. Alterations in expression of OATPs will cause altered uptake of the contrast agent, while the intracellular concentration of Gd-EOB-DTPA after it enters the hepatocyte is strongly influenced by the ability of MRPs to excrete the contrast agent into the bile ducts or back into the sinusoids. Gd-BOPTA shares several properties with Gd-EOB-DTPA, including initial distribution in the extracellular fluid compartment and being uptaken via organic transporter on hepatocytes. Therefore, both can be used as an ECF agent for hepatic arterial and portal venous phase imaging and as a hepatobiliary agent for delayed hepatocellular phase imaging. While Gd-BOPTA is administered at a 4-fold higher dose than Gd-EOB-DTPA, only 3-5% of Gd-BOPTA is cleared via biliary excretion, significantly lower than 50% for Gd-EOB-DTPA. Because of its greater hepatic uptake, Gd-EOB-DTPA results in greater hepatocellular phase enhancement than Gd-BOPTA and is thus a preferred hepatobiliary agent. In addition, Gd-EOB-DTPA is cleared from the body more rapidly, according to the elimination half-lives (1 hour for Gd-EOB-DTPA and 1–2 hours for Gd-BOPTA).

Figure 1. Uptake and excretion mechanism of hepatobiliary agent Gd-EOB-DTPA and Gd-EOB-DTPA-enhanced MRI for assessment of hepatocyte transport function.

(A) Diagram shows the hepatocyte uptake and biliary excretion mechanism of Gd-EOB-DTPA. Axial T1-weighted fat saturated 3D gradient echo images show hepatobiliary phase of enhancement imaged at 20 minutes post administration for Gd-EOB-DTPA for (B) a healthy subject and (C) a patient with NASH induced liver cirrhosis. The biliary ducts filled by Gd-EOB-DTPA are indicated by white arrowheads. The morphologic stigmata of cirrhosis such as hypertrophy of the caudate lobe and left lobe of the liver are indicated by yellow arrows. (D) Hepatocyte uptake ratio map in a 57-year-old man with hepatitis B cirrhosis and Child-Pugh class A disease reveals a hepatocyte uptake ratio value of 3.64; indocyanine green retention test is 14.6%. (E) Hepatocyte uptake ratio map in a 55-year-old man with hepatitis B cirrhosis and Child-Pugh class A disease reveals a hepatocyte uptake ratio value of 1.58; indocyanine green retention test is 22.9%. (Figure 1A created with BioRender.com; Figure 1D and 1E adapted from [75])

Superparamagnetic iron oxide particles

SPIO particles are composed of small crystalline cores of iron oxides coated with a biocompatible material, e.g. dextran or citrate, to improve solubility and tolerability. Typically given as a slow infusion, SPIOs are phagocytosed by macrophages throughout the body, and especially by Kupffer cells which line the hepatic sinusoids [47]. SPIO contrast agents can be classified in two groups, superparamagnetic iron oxide particles (SPIO, <150 nm in diameter) and ultra-small superparamagnetic iron oxide particles (USPIO, <30 nm in diameter) [32, 48, 49]. Typically, SPIOs of > 80nm diameter are rapidly taken up by the RES while smaller USPIOs can evade macrophage capture and exhibit a longer blood circulation time of hours to days. While several SPIOs have been approved as liver specific contrast agents, they have since been withdrawn from one or all major markets due to poor sales. The USPIO ferumoxytol is currently available as an iron replacement therapy but it sees some off-label use as a contrast agent. Compared to liver-specific SPIOs used previously, uptake of ferumoxytol into Kuppfer cells is slow.

Molecularly targeted agents

Molecularly targeted probes are small molecules, peptides, or antibodies that recognize a specific protein, receptor, or biological process. They are tagged with a contrast generating moiety, such as chelated gadolinium for MRI visualization. There has been considerable effort to develop targeted MRI probes for noninvasive imaging of fibrotic or inflammatory molecular pathways, common features shared by almost all causes of progressive CLDs. While still under preclinical development, targeted probes provide unique means to characterize and quantify dysregulated molecular events, playing an important role in optimizing the drug discovery and validation processes. Their clinical translation, if successful, can profoundly impact the detection, staging, prognosis, and treatment monitoring of disease, as well as elucidating complex biology.

Functional MRI of diffuse liver diseases

Perfusion imaging

Dynamic contrast enhanced (DCE) MRI with the ECF agents has become an integral part of clinical liver MRI to inform on changes in tissue perfusion. A volumetric T1-weighted fat-suppressed gradient echo technique is commonly used to achieve high spatial resolution and sufficient temporal resolution in DCE MRI. To ensure alignment of the liver in serial images, motion control is achieved either prospectively using respiratory control techniques, e.g. multiple breathholds, navigated acquisition, and respiratory triggering, or retrospectively by image registration [50, 51]. Fat suppression is necessary to reduce the high T1-weighted signal from subcutaneous and juxtadiaphragmatic fat stores that is responsible for chemical shift artifact and is often visible as motion artifact [52].

Perfusion MR parameters can be derived using model-free or model-based techniques [53]. Model-free analysis refers to the parameters extracted from time-signal intensity curves, including area under curve (AUC), time to peak, peak enhancement, wash-in slope, etc. The AUC over 60-120 seconds from the onset of contrast enhancement is a widely used parameter, which represents the leakage of contrast into the extracellular space and therefore, indirectly reflects tissue perfusion. Model-based approaches involve curve fitting of a dual-input single-compartmental model [54, 55]. The model-derived parameters quantify the pharmacokinetic distribution of the agent in physiological terms like arterial and portal venous blood flow, distribution volume, and mean transit time, although it should be noted that several assumptions underlie the model-based approaches which can affect the accuracy of the extracted parameters.

Quantitative DCE MRI was used in research trials to assess the microcirculatory changes in liver fibrosis and cirrhosis. The deposition of collagen in the space of Disse and sinusoidal capillarization results in an increased resistance to incoming sinusoidal blood flow [56], which leads to a decrease in portal venous flow to the liver, an increase in hepatic arterial flow, and the subsequent formation of intrahepatic and portosystemic shunts [57]. Transfer of the ECF agent from the vascular sinusoids into the interstitial space is thus increasingly impeded by liver fibrosis [53]. In patients with cirrhosis, decreased portal and total hepatic perfusion is observed, as well as increased arterial perfusion and mean transit time, with preserved or increased distribution volume. These perfusion changes occur already at intermediate stages of liver fibrosis, but are more marked in cirrhosis, where they correlate with the degree of liver dysfunction and portal hypertension [58]. A dual-input, single-compartment, model-based study reported an increase in absolute arterial blood flow, arterial fraction, distribution volume, and mean transit time and a decrease in portal venous fraction, in patients with advanced liver fibrosis compared with patients with early-stage liver fibrosis [59]. Among all the estimated perfusion parameters, the distribution volume had the best performance in predicting advanced liver fibrosis, with a sensitivity of 77% and a specificity of 79%.

Imaging hepatocyte function

Active transport of a hepatobiliary contrast agent into the hepatocytes offers a unique means for functional assessment of the liver. In addition to the initial dynamic phase, a hepatobiliary phase (HBP) of enhancement is imaged at about 20 minutes post contrast administration for Gd-EOB-DTPA and at 1-2 hours post Gd-BOPTA. During the HBP, only the liver parenchyma but not the vessels is enhanced, while impaired hepatocytes commonly show hypointensity [41]. Signal enhancement in HBP images reflects the balance between uptake and excretory transporters. It also reflects the total number of hepatocytes; for instance in fibrosis and cirrhosis decreased relative liver enhancement (RLE) results from the loss of some normal hepatocytes due to fibrotic replacement. Based upon a variety of pharmacokinetic properties, including liver perfusion, vascular permeability, extracellular diffusion, and hepatocyte transporter expression, the hepatobiliary contrast agents, particularly Gd-EOB-DTPA provide a combination of morphologic and functional information simultaneously in hepatobiliary systems [41, 60, 61].

Typically, liver function assessment is performed using the signal enhancement on the HBP images acquired at 20 min post Gd-EOB-DTPA, with some studies using the indocyanine green retention test (ICG-R15) or clinical liver function for comparison [29, 41, 62-64]. However a simple signal intensity measurement at a single timepoint may be subject to technical variations between scanners as well as physiological variations between healthy subjects and patients. Moreover, strong functional heterogeneity exists, particularly in diseased livers, which cannot be captured by region-of-interest based signal intensity [65]. Therefore, dynamic image acquisition is increasingly employed to enable pharmacokinetic analysis for extracting both temporal and spatial information. The feasibility to derive hepatic extraction fraction (HEF) from DCE MRI with Gd-EOB-DTPA by deconvolutional analysis was demonstrated early on in healthy subjects [66]. In a rabbit model of carbon tetrachloride (CCl₄)–induced liver fibrosis, the HEF was calculated by deconvoluting aortic and hepatic parenchymal time-intensity curves from DCE MRI of Gd-EOB-DTPA [67]. A negative correlation was found between the HEF and changes in ICG R15. A multicenter study determined liver function using DCE MRI of Gd-EOB-DTPA in patients scheduled to undergo hepatectomy or radiofrequency ablation for hepatocellular carcinoma [68]. The predicted remnant liver function, calculated as the pre-treatment HEF multiplied by the remnant liver volume, showed a significant negative correlation with post-treatment ICG-R15. In contrast to global liver function tests such as ICG-R15, DCE-MRI with Gd-EOB-DTPA demonstrated the potential to include regional variations of liver function into planning a personalized preoperative strategy. With pharmacokinetic analysis of DCE MRI data, liver perfusion and hepatocyte transport function can be assessed separately using deconvolutional or compartmental analysis [69, 70]. While these quantitative methods are promising, standardization on processing and analysis will be needed to obtain clinically useful metrics.

Gd-EOB-DTPA enhanced MRI has been used to assess hepatobiliary function in liver fibrosis, cirrhosis, and non-alcoholic steatohepatitis (NASH) [71-74]. Axial T1-weighted fat saturated 3D gradient echo images obtained 20 minutes after injection of Gd-EOB-DTPA are shown in a volunteer with normal liver function (Figure 1B) and in a patient suffering from NASH induced cirrhosis (Figure 1C). The biliary ducts filled by Gd-EOB-DTPA are indicated by arrowheads. The liver parenchyma demonstrates marked and homogenously increased signal intensity in the healthy liver and only mild signal enhancement in the cirrhotic liver. The morphologic stigmata of cirrhosis such as hypertrophy of the caudate lobe and left lobe of the liver are also seen (yellow arrows). In an animal model of advanced liver fibrosis, the enhancement parameters (maximum enhancement, and time to peak) and the pharmacokinetic parameters (HEF and mean residence time) measured from DCE MRI significantly decreased in diseased rats relative to those in control rats, which correlated with the expression of the hepatic organic anion transporters [71]. In a clinical study in patients with CLD or cirrhosis, Gd-EOB-DTPA signal enhancement in the liver was normalized to that in the spleen to quantify the hepatocyte uptake ratio [75]. Hepatocyte uptake ratios are negatively correlated with liver function measured by ICG-R15. In patients with CLD or Child-Pugh class A, significantly higher hepatocyte uptake ratio was found for those with ICG-R15 ≤ 20% (Figure 1D) than those with ICG-R15 ≥ 20% (Figure 1E). Hepatocyte uptake ratios demonstrated better performance for detecting ICG-R15 ≥ 20% with the highest area under the receiver operating characteristic curve (AUROC) (0.96; 95% confidence interval: 0.89 - 0.99), followed by postcontrast liver T1 value (0.89; 0.76 - 0.97) and liver volume (0.70; 0.55 - 0.83). On the other hand, impaired hepatic uptake of Gd-EOB-DTPA was found to correlate with fibrosis stage in patients with liver fibrosis/cirrhosis [20, 74, 76, 77]. Recently, a semi-quantitative functional liver imaging score (FLIS) derived from the HBP image of Gd-EOB-DTPA was used to predict hepatic decompensation and transplant-free survival in patients with CLD [78]. Because FLIS requires no dynamic acquisition or signal modeling and is independent of MRI field-strength and vendor, it may be easier to implement in routine clinical practice.

In the past decade, Gd-EOB-DTPA–enhanced MRI was increasingly used to assess risk of post-hepatectomy liver failure and graft liver dysfunction. An accurate estimate of reserve hepatic function before resection of diseased liver is crucial to avoid post-hepatectomy liver failure [79]. Several clinical studies showed that lower preoperative Gd-EOB-DTPA uptake in the liver was associated with a higher probability of post-hepatectomy liver failure [80-82]. Orthotopic liver transplantation is the only curative therapeutic option for end-stage liver diseases, and prompt diagnosis and aggressive management of possible complications are critically needed to give each graft the best chance of survival. Gd-EOB-DTPA–enhanced MRI was used to evaluate global and regional liver function in liver transplant recipients with regard to serum biomarkers and mortality at 1 year from imaging [83]. Decreased hepatic uptake and delayed or absent excretion of Gd-EOB-DTPA, in the absence of biliary obstruction, was found to correlate with inferior short-term graft survival, whereas a normal hepatobiliary excretion was associated with 1-year retransplantation-free survival [83]. The same group described a qualitative scoring system based on three simple features derived from Gd-EOB-DTPA–enhanced MRI to predict graft survival for three years [84].

In patients with end-stage renal failure, liver enhancement in HBP was markedly reduced, attributed to elevated serum ferritin levels. Increased hepatic accumulation of iron, given its superparamagnetic properties, reduces the signal enhancement induced by Gd chelates. Therefore, Gd-EOB-DTPA is not administrated in patients with elevated serum ferritin (>200 ng/mL in men and >150 ng/mL in women recommended by WHO) or total serum bilirubin >3 mg/dL per drug’s prescribing information due to inadequate liver enhancement.

Compared to Gd-EOB-DTPA, a much lower biliary excretion and a higher typical dose of Gd-BOPTA result in higher enhancement of the hepatic arterial, portal and hepatic venous branches during dynamic liver imaging, but also in a delayed onset of the HBP (1-2 hours versus 20 minutes for Gd-EOB-DTPA) and in a relatively weaker HBP signal enhancement. Similar to Gd-EOB-DTPA, the RLE after Gd-BOPTA might provide insights into liver function. RLE was inversely correlated with prothrombin time and total bilirubin levels and directly correlated with serum albumin levels, helping distinguish Child-Pugh A from Child-Pugh B patients [85]. However, the late HBP and varied excretion time of Gd-BOPTA makes it less suitable for assessing liver function.

Double contrast-enhanced MRI of liver fibrosis

Fibrotic tissue has reduced Kupffer cell density and less accumulation of iron oxides. Therefore, SPIO-enhanced MRI shows liver fibrosis as hyperintense reticulations with the background liver darkened preferentially [86]. SPIOs can be combined with ECF agents for double contrast-enhanced MRI [87, 88]. The SPIO produces a dark liver background, while the gadolinium-based ECF agent provides a delayed enhancement of septal and bridging fibrosis. The two agents synergistically improve fibrosis-to-normal liver contrast with higher clarity than can be achieved with either agent alone [89, 90]. Aguirre et al. showed that the hepatic texture alterations seen on double contrast-enhanced MRI differentiated advanced hepatic fibrosis from mild or absent fibrosis with an accuracy of 93% [89]. The authors suggested that combining texture, surface, and secondary imaging features would probably further improve diagnostic performance. The main limitations of double contrast-enhanced MRI are the higher cost and inconvenience associated with dual contrast administration, as well as the limited availability of SPIOs.

Imaging molecular processes involved in liver fibrosis

Liver fibrosis results from repeated hepatic injury which causes the chronic activation of tissue repair mechanisms to replace necrotic tissue with extracellular matrix (ECM) proteins [91]. Figure 2 illustrates the cellular and molecular processes involved in the pathogenesis of liver fibrosis. First, inflammatory responses to the injury occur, which involve the recruitment and stimulation of immune cells to the site of injury, as well as the stimulation of Kupffer cells. Loss of endothelial fenestrations allows inflammatory immune cells to infiltrate into the hepatic parenchyma along with plasma proteins like fibrinogen. Thrombin mediated extravascular coagulation results in fibrin deposition. Soluble inflammatory mediators including chemokines and cytokines released by the inflammatory cells activate the quiescent hepatic stellate cells (HSCs). Activated HSCs transdifferentiate to myofibroblasts and produce and deposit collagen. Activated HSCs and other matrix-producing cells including bone marrow-derived fibrocytes and portal fibroblasts, drive enhanced expression, secretion, and deposition of ECM.

Figure 2. Cellular and molecular mechanisms in the pathogenesis of liver fibrosis.

Injury to hepatocytes results in the recruitment and stimulation of inflammatory immune cells to the site of injury, as well as the stimulation of resident macrophages, i.e. Kupffer cells. The loss of endothelial fenestrations allows inflammatory cells and plasma proteins like fibrinogen to infiltrate in the hepatic parenchyma. Thrombin-driven hepatic fibrin deposition is also frequently observed in the parenchyma. Soluble mediators including chemokines and cytokines released by the inflammatory cells activate the quiescent hepatic stellate cells (HSCs). The activated HSCs transdifferentiate to myofibroblasts, which are the predominant source of collagen synthesis and deposition. Activated HSCs, complemented by other matrix-producing cells including bone marrow-derived fibrocytes and portal fibroblasts, lead to enhanced expression and secretion of extracellular matrix and matrix deposition. (Illustration created with BioRender.com)

Imaging matrix deposition and cross-linking

Excess deposition of type I collagen is the hallmark of liver fibrosis. EP-3533 is a molecular probe that contains a cyclic peptide specific to type I collagen and 3 Gd-DTPA moieties to boost MR signal [92, 93]. Compared to the ECF agent Gd-DTPA, EP-3533–enhanced MRI could distinguish fibrotic liver from controls in a rat model of diethylnitrosamine (DEN)-induced fibrosis and a mouse model of CCl₄-induced liver fibrosis [94]. The change in MRI liver-to-muscle contrast correlated linearly to collagen content determined biochemically by hydroxyproline analysis and to the Ishak score of fibrosis. In CCl₄-injured mice, rats with bile duct ligation (BDL), and DEN-injured rats, EP-3533 was shown to accurately stage liver fibrosis [95-97]. EP-3533–enhanced MRI could also assess treatment response, which historically has relied on histologic analysis [98, 99]. In a rat BDL model, EP-3533–enhanced MRI detected significant changes in fibrosis in response to rapamycin therapy and could prospectively identify treatment responders from non-responders [96]. Moreover, 3D molecular MRI enabled characterization of intrahepatic heterogeneity throughout the whole liver in a rapamycin non-responder rat, in which the fibrosis heterogeneity was confirmed by histology (Figure 3A-C). EP-3533–enhanced MRI also reported response to the farnesoid X receptor agonist EDP-305 in the form of decreased fibrosis in BDL rats [100]. To improve the translational potential of EP-3533, a new version, called CM-101, employing the highly stable macrocyclic Gd-DOTA chelate was developed, and was shown to detect fibrosis in rat BDL and mouse CCl₄ models with faster rapid blood clearance and minimal accumulation of gadolinium in bone or tissue compared to EP-3533 [101].

Figure 3. Type I collagen imaging in murine models of liver fibrosis.

(A) Pre- and post-EP-3533 longitudinal relaxation rate (R1) maps of a bile duct ligated (BDL) rat treated with rapamycin. Striking heterogeneity in liver fibrosis with greater R1 in the right liver lobe compared to the left liver lobe. (B) While the pre-EP-3533 histogram (blue) of liver R1 values was homogeneous the post EP-3533 R1 histogram (red) demonstrated a bimodal R1 distribution. (C) Corresponding Sirius Red images from the left and right liver lobes confirmed the fibrosis heterogeneity. ((D) R1 maps of normal (Ishak stage), early-stage (Ishak stage 3), and late-stage (Ishak stage 5) liver fibrosis before and 24 h after injection of ProCA32.collagen1. (E) αSMA, H&E, and Sirius red staining of early- and late-stage fibrotic liver compared to normal liver confirms the stage of liver fibrosis. (Figure 3A-C and captions adapted from [96], figure 3D-E and captions adapted from [102])

ProCA32 is a small protein that was engineered to bind tightly to gadolinium and to have a very high relaxivity. Grafting the same type I collagen targeting peptide used in EP-3533 to ProCA32 created ProCA32.collagen1 and this reagent was evaluated in alcohol/chemical-induced mouse models of liver fibrosis and a diet-induced mouse model of NASH [102]. ProCA32.collagen1–enhanced MRI could detect early-stage alcohol-induced liver fibrosis (Ishak stage 3) and early-stage NASH (Ishak stage 1) using both longitudinal and transverse relaxation (R1 and R2) maps at 24h post-injection. R1 quantification of ProCA32.collagen1 could distinguish early-stage (Ishak stage 3) from late-stage fibrosis (Ishak stage 5), which correlated with histologic quantification of fibrosis (Figure 3D-E). Because of its much larger size, ProCA32.collagen1 has a longer blood residency time than either EP-3533 or CM-101. Therefore the delayed imaging took place hours after injection or the next day compared to within 10 – 40 minutes with EP-3533 or CM-101.

Elastin accumulates at advanced stages of liver fibrosis due to both increased synthesis and decreased macrophage metalloelastase (MMP12)-mediated degradation [103, 104]. The gadolinium-based elastin-specific MR probe, Gd-ESMA, was used to monitor ECM-remodeling during liver fibrosis [105]. Gd-ESMA administration resulted in significant perivascular signal enhancement in the livers of CCl₄ treated mice, but not in control mice. The periportal elastin deposition detected by Gd-ESMA-enhanced MRI was further confirmed by elastin-specific Elastica-Van-Gieson staining.

Fibrin-fibronectin complexes deposit in fibrotic liver as a result of cross-linking between fibrin/fibrinogen and fibronectin [106], and can serve as a molecular target for liver fibrosis. Fibronectin deposition in the liver was evaluated using Gd-P, a CGLIIQKNEC (CLT1) peptide-targeted nanoglobular probe that binds to fibrin-fibronectin complexes. Using a CCl₄ liver injury mouse model, an increased amount of fibronectin in the extracellular space in insulted livers was confirmed by histology, and Gd-P induced signal enhancement was significantly higher in injured mice than controls [107].

Imaging fibrogenesis

Fibrogenesis, or the active deposition of ECM, is characterized by high levels of allysine, a reactive aldehyde generated as part of a lysyl oxidase (LOX)-mediated collagen cross-linking process. Allysine-targeting imaging probes can be used to monitor fibrogenesis [108, 109]. The hydrazide containing, aldehyde targeting agent Gd-Hyd detected and staged disease progression caused by CCl₄ injury in mice and reported on reduced fibrogenesis in mice after withdrawal of CCl₄ as assessed by histology and lysyl oxidase gene expression [110]. In mice fed with a choline-deficient, L-amino acid-defined, high-fat diet (CDAHFD) model of NASH, Gd-Hyd detected reduced fibrogenesis after treatment with the farnesoid X receptor agonist EDP-305, corroborated by picrosirius red collagen staining, hydroxyproline biochemical quantification, lysyl oxidase gene expression analyses [100]. These data suggest that Gd-Hyd could be used to monitor disease progression and regression. The fibrogenesis probe shows potential for predicting treatment responses, where its performance in comparison with that of serum-based markers warrants further evaluation.

Activated HSCs express integrin αvβ3 [111, 112]. ECM proteins bind to integrins like αvβ3 by means of the tri-amino acid sequence arginine-glycine-aspartate (RGD) [113, 114]. Numerous RGD-based integrin-binding agents have been developed for imaging αvβ3 in activated HSCs [115-119]. For example, an RGD peptide-based USPIO, a T2 MRI contrast agent, was developed to image the activated HSCs in CCl₄ injured rats [118]. The integrin-targeted iron particles were engulfed by activated HSCs, resulting in reduction of T2 relaxation times. However, the USPIOs were also taken up by Kupffer cells in both normal and injured livers resulting in a non-specific background.

Imaging the inflammatory response

Tissue injury triggers extravascular coagulation as a part of inflammatory response and fibroproliferative processes of tissue repair [120], and results in fibrin deposition [121]. Abnormal extravascular deposition of fibrin was detected in rodent models of liver injury [106, 122]. EP-2104R is a fibrin-specific, peptide-based, Gd probe that has been used for molecular imaging of extravascular fibrin in mouse models of cancer and pulmonary fibrosis [123-125]. In DEN-injured rats signal enhancement with EP-2104R was significantly higher in animals imaged at 1 day post-DEN dosing when liver inflammation was high compared with 7 days post-DEN when inflammation had subsided as confirmed by histology (Figure 4A-B) [126]. Collagen imaging with EP-3533 showed equivalent T1 change when imaging rats 1 day or 7 days post-DEN, consistent with equivalent fibrosis assessed by Ishak score, demonstrating that fibrin/inflammation could be imaged on a background of liver fibrosis.

Figure 4. Molecular MRI of inflammatory responses.

(A) Molecular MRI using the fibrin-specific MRI probe EP-2104R for the noninvasive imaging of fibrin as a marker of liver inflammation in diethylnitrosamine (DEN)-injured liver. Representative T1-weighted axial images of rats dosed weekly for 4 weeks with PBS (top), or with 100 mg/kg of DEN and imaged 1 day after the last DEN dose (bottom). False color overlay represents subtraction image of precontrast image from the 1-minute post EP-2104R image. (B) Histological analysis of liver specimens from rats received PBS (top) or DEN (bottom). Hematoxylin and eosin staining of liver tissue showed inflammatory regions labeled with pointed black arrows. Sirius Red staining of liver tissue showed fibrosis by intense red staining of collagen fibers. Immunohistochemistry staining of fibrin in liver tissue. (C) Molecular MPO-Gd–enhanced MRI depicting inflammatory activity in liver biopsy samples from NASH (top) and control patients (bottom). Pseudo-colored MPO-Gd–enhanced MR images are shown with the biopsy sample outlined by dashed line. Increased signal after incubation with MPO-Gd is seen on samples from NASH patients, but not control patients. (D) Immunohistochemistry for MPO in liver biopsy samples from NASH (top) and control patients (bottom) (bar = 100 μm) demonstrated clusters of MPO-expressing cells in samples from NASH patients but not in samples from control patients. (Figure 4A-B and captions adapted from [126]; figure4C-D and captions adapted from [140])

The inflammatory response also results in recruitment of immune cells including monocytes and neutrophils to the site of injury [127, 128]. These pro-inflammatory immune cells are abundant in myeloperoxidase (MPO), a heme-containing enzyme capable of generating reactive oxygen and nitrogen species (ROS and RNS) for host defense [129]. Increased MPO is found in human livers with NASH compared with steatotic specimens [130]. Inflammation and fibrosis were also decreased in MPO-deficient mice fed with a high fat diet [131], indicating the possible role of MPO in the pathogenesis of NASH. Moreover, MPO may be important in the development of liver fibrosis through activation of HSCs [132]. An activatable probe, MPO-Gd, consists of a Gd-chelate bearing two 5-hydroxytryptamide (5-HT) moieties that are readily oxidized by MPO [133, 134]. Oxidation of MPO-Gd leads to its oligomerization and/or covalent addition to proteins; both processes increase relaxivity and help retain the probe at the site of injury, thereby allowing for noninvasive assessment of MPO activity [135, 136]. MPO-Gd was used for in vivo detection of MPO activity in animal models of myocardial infarction, multiple sclerosis, stroke, and vasculitis [135, 137-141]. In a study using two mouse models of steatohepatitis, elevated MPO activity could be detected with MPO-Gd, allowing differentiation of steatohepatitis from steatosis [142]. Ex vivo molecular MRI of liver biopsy samples from NASH and control patients confirmed results from animal studies (Figure 4C-D). Improved MPO-targeted probes using the more stable and translatable macrocyclic Gd-DOTA have been reported [143, 144]. Recently a redox-active iron complex, Fe-PyC3A, was shown to detect tissue inflammation and in vivo Fe-PyC3A-induced signal enhancement correlated strongly with ex vivo quantitation of MPO activity [145].

Multiparametric imaging of diffuse liver diseases

In addition to functional and molecular liver MRI, state-of-the-art MRI techniques in current practice and in development for noninvasive assessment of CLD include MR elastography, diffusion-weighted imaging, MRI quantification of liver fat or iron content. Molecular and/or functional imaging can be readily added to these advanced MRI protocols to provide a more comprehensive assessment of CLD. For instance Gd-Hyd–enhanced MRI of fibrogenesis and Dixon MRI-based quantitative fat imaging were used in a mouse model of NASH [100]. In a DEN-induced rat model of fibrosis, collagen imaging with EP-3533 was shown to be most sensitive to early fibrosis, while tissue stiffness assessed by MR elastography was more sensitive to advanced fibrosis [97]. A composite score formulated from these two measurements resulted in increased diagnostic accuracy for staging fibrosis [97]. Another multi-parametric study with diffusion-weighted imaging, susceptibility-weighted imaging, and Gd-EOB-DTPA-enhanced MRI showed improved performance in staging fibrosis [146]. Multi-parametric approaches combining native T1 mapping for fibrosis/inflammation staging, T2* mapping for liver iron quantification, and proton MR spectroscopy for liver fat quantification have been trialed to detect liver fibrosis, steatosis, and hemosiderosis, respectively [147, 148].

Recently a multi-parametric MRI study, including type-1-collagen-specific probe EP-3533, allysine-targeted fibrogenesis probe Gd-Hyd, MR elastography, and native T1, was used to characterize fibrosis and to assess treatment response in rats fed with CDAHFD as a model of NASH [149]. The collagen-targeted molecular probe most accurately detected early onset of liver fibrosis with an AUROC of 0.95, superior to other MRI measures. The fibrogenesis imaging probe Gd-Hyd had the highest accuracy in detecting responder and non-responders in treatment groups. The advanced MRI techniques applied there are complementary in their capture of the fibrotic process. All techniques were performed in a single protocol, highlighting the strength of MRI for characterizing liver disease in the context of NASH. Despite higher cost, the rich information provided by such imaging protocol raises new possibilities for clinical imaging and for assessing treatment response for new therapies in development.

Challenges and Future Directions

Several challenges exist in developing and translating molecular MRI probes. First, newly developed contrast agents are considered novel drugs and therefore require extensive preclinical safety evaluations, and manufacturing must be performed to meet good manufacturing practice standards, both of which create a large cost barrier for first-in-human evaluation. Second, animal models may not reflect target concentrations in human disease, and probe pharmacokinetics and metabolism may also differ. Furthermore, quantification of sensitivity and specificity will require an effective truth standard like histology, which may not be readily available. Quantification also requires rigorous assessment of reproducibility among subjects and different vendors’ scanners. Ultimately outcome data will be required to establish the clinical utility of the probes.

However in the era of precision medicine and targeted therapy the need for noninvasive quantification of liver function and visualization of cellular/molecular processes underlying liver diseases is continuously growing. With the development of expensive new anti-inflammatory/antifibrotic therapeutics, imaging could be deployed to stratify patients to beneficial therapies and for monitoring treatment response. The functional and molecular imaging approaches described here represent the first steps towards these goals. While much of the work was performed in preclinical models or a small cohort of patients with single disease condition, the next steps will hinge upon improved and standardized dynamic MRI and post-processing methods, rational design and optimization of the molecular probes, and further prospective validation studies, for successful translation into clinical application.

Conclusion

Advances in MRI, particularly with the wide deployment of hepatobiliary-specific contrast agents and the development of molecularly targeted probes, have provided the opportunity to revolutionize how we evaluate various liver pathologic conditions. Despite the challenges to establish their clinical usage, we anticipate the coalescence of major advances in engineering, molecular biology, chemistry, immunology, and genetics will continue to fuel multi-disciplinary innovations and drive the field of clinical noninvasive imaging that will ultimately allow disease identification, risk stratification, and monitoring of therapy effects with unparalleled sensitivity and specificity. Moreover, techniques that allow imaging of molecular and cellular events go hand in hand with the development of molecular therapies, offering promise for successfully combining imaging with therapy.

Key points.

• The value of MRI in the liver is significantly enhanced by a wide range of contrast agents, both clinically available and under development, that increase the distinction between normal and pathological tissues and can assess liver function.

• Dynamic contrast-enhanced MRI with conventional gadolinium-based extracellular fluid intravenous contrast agents is used to detect changes in perfusion, blood flow, and vascularity associated with liver lesions, cirrhosis or chronic liver disease.

• In contrast to global liver function tests such as ICG-R15, dynamic contrast-enhanced MRI with hepatocyte-specific contrast agents such as Gd-EOB-DTPA offers regional information about hepatocyte transport function during the hepatobiliary phase, improving the assessment of diseased hepatobiliary systems.

• Liver perfusion and hepatocyte transport function can be assessed separately using pharmacokinetic analysis of dynamic Gd-EOB-DTPA-enhanced MRI data.

• Liver fibrosis is a common feature of almost all causes of progressive chronic liver disease, which is associated with a cascade of cellular and molecular processes such as immune system activation, vascular leak, extravascular coagulation, collagen deposition coagulation, inflammation, excessive deposition of extracellular matrix and cross-linking.

• Novel molecular contrast agents that are biochemically targeted to specific cellular/molecular processes enable staging liver fibrosis or detecting treatment responses, and are undergoing preclinical development.

• Given the multiparametric capabilities of MRI, functional and molecular MRI can be readily performed in conjunction with advanced MRI techniques such as MR elastography, diffusion-weighted imaging, MRI quantification of liver fat or iron content to provide a more comprehensive characterization of liver diseases and to assess their response to treatment.