Journal of Hepatology
Volume 40, Issue 1 , Pages 162-171, January 2004

Magnetic resonance imaging of liver tumors

Division of Diagnostic and Interventional Radiology, Department of Oncology, Transplants, and Advanced Technologies in Medicine, University of Pisa, Via Roma 67, I-56126 Pisa, Italy

Article Outline

 

Back to Article Outline

1. Introduction 

Over the past few years, magnetic resonance imaging (MRI) of the liver has progressed significantly. Technical advances in hardware and software have allowed the acquisition of images with excellent anatomic detail, largely free of artifacts secondary to respiratory motion. Fast sequences have reduced image acquisition time, thereby improving patient acceptance and allowing more efficient utilization of machine time. New volumetric sequences have enabled three-dimensional serial dynamic imaging of the liver with a very high spatial and temporal resolution, reducing section misregistration and motion artifacts while improving multiplanar reformations [1], [2]. A number of novel tissue-specific contrast agents, including hepatocyte-targeted and reticulo-endothelial system (RES)-targeted compounds, have been developed, permitting manipulation of tissue signal in different ways, according to the relevant diagnostic issue [3], [4]. MR angiography, MR venography, and MR cholangiography have been introduced in clinical practice, allowing a comprehensive assessment of parenchymal, vascular, and biliary structures in one examination. This article reviews the current status of MRI for liver tumor imaging, discussing technique and methodology, use of contrast agents, and main clinical applications.

Back to Article Outline

2. Technique and methodology 

A careful selection of imaging strategies can currently yield a comprehensive assessment of the liver in a reasonably short examination time, thereby increasing patient throughput and decreasing study cost [5]. The implementation of high-performance gradient systems allows for rapid acquisition times that can bypass many of the motion artifacts that previously posed limitations to liver MRI studies. T1-weighted images are acquired by using multisection spoiled gradient-echo sequences, with or without fat suppression, that allow thin-section imaging of the entire liver during a single suspended respiration. If volumetric three-dimensional methods are used, images may be obtained in all the planes, enabling improved detection, localization, and characterization of small lesions. T2-weighted images are usually obtained through breathing-averaged or respiratory-triggered T2-weighted fast spin-echo sequences with fat saturation. T2-weighted breath-hold imaging is also possible by using single-shot or fast-recovery fast spin-echo sequences [6]. Additional sequences, such as out-of-phase spoiled gradient-echo T1-weighted sequences, may be performed to provide a comprehensive information or to solve specific diagnostic issues. After the baseline study, contrast-enhanced sequences are almost routinely obtained. The contrast-enhanced study is currently a fundamental component of an MRI examination of the liver in most instances. Also, a comprehensive MRI study of liver may include, when indicated, MR angiography, MR venography, and MR cholangiography [7].

Back to Article Outline

3. Contrast agents for liver MRI 

Contrast agents have become an essential part of a liver MRI examination in a number of circumstances. They can act as positive or negative enhancers, depending on their nature, concentration, and the type and timing of imaging sequences. The contrast agents available for hepatic imaging can be divided into three categories, according to their biodistribution: (a) extracellular fluid space contrast agents; (b) hepatocyte-targeted contrast agents; and (c) RES-targeted contrast agents (Table 1). There is a wide variation among these compounds with regard to contrast behavior and tissue specificity.

Table 1. Classification of contrast agents for magnetic resonance imaging of the liver
Category and compoundsImaging properties and main indication
Extracellular fluid space contrast agents
– Gd-DTPA (Magnevist®, Schering AG)– T1 positive enhancement
– Gd-DTPA-BMA (Omniscan®, Amersham Health)– Standard compounds for multiphase dynamic imaging
– Gd-HP-DO3A (ProHance®, Bracco)– Lesion detection (hypervascular tumors)
– Gd-DOTA (Dotarem®, Laboratoires Guerbet)– Lesion characterization
Hepatocyte-targeted contrast agents
– Gd-BOPTA (MultiHance®, Bracco)a– T1 positive enhancement
– MnDPDP (Teslascan®, Amersham Health)– Lesion detection (hypovascular tumors)
– Gd-EOB-DTPA (Schering AG)a,b– Lesion characterization (hepatocellular tumors)
RES-targeted contrast agents
– AMI-25 (Endorem®, Laboratoires Guerbet)– T1 positive enhancement, T2 negative enhancement
– SH U 555 A (Resovist®, Schering AG)a– Lesion detection (hypovascular tumors)
– Lesion characterization

Note. RES, reticulo-endothelial system.

a Suitable for bolus injection and dynamic post-contrast imaging.

b Not yet approved for clinical use in Europe.

3.1. Extracellular fluid space contrast agents 

Extracellular fluid space contrast agents are hydrophilic, small-molecular-weight gadolinium chelates. These agents are non-specific compounds that, following intravenous injection, are rapidly distributed from the vascular compartment to the interstitial spaces and then excreted via the urinary system. They show a paramagnetic effect and, hence, they reduce T1 relaxivity to a much greater extent than T2 relaxivity, thus increasing signal intensity on T1 weighted images. Gadopentetate dimeglumine (Gd-DTPA, Magnevist®, Schering AG) has been the first MRI contrast agent approved for clinical use. Other non-specific gadolinium complexes which are currently available on the market include gadodiamide (Gd-DTPA-BMA, Omniscan®, Amersham Health), gadoteridol (Gd-HP-DO3A, ProHance®, Bracco), and gadoterate meglumine (Gd-DOTA, Dotarem®, Laboratoires Guerbet). All these agents exhibit a well-established record of safety [8].

Extracellular fluid space contrast agents provide information on vascularization and perfusion similar – to some extent – to that of iodinated contrast agents for computed tomography (CT). Following the introduction of breath-hold T1-weighted sequences, that can be repeated in a serial dynamic fashion after bolus injection of extracellular gadolinium chelates, the clinical usefulness of these agents has greatly increased [9], [10]. They have rapidly become an essential part of a comprehensive MR examination of the liver in a number of applications. A dynamic contrast-enhanced study of the liver is performed with a timing scheme that enables selective imaging during the arterial phase, the portal venous phase, and the delayed phase of enhancement. The most popular technique for dynamic studies is the spoiled gradient-echo technique, that allows thin-section imaging of the entire liver during a single breath-hold. If volumetric three-dimensional sequences are used, the volumetric data sets can be reconstructed in any plane, enabling improved detection, localization, and characterization of small liver lesions. The combination of high-resolution three-dimensional sequences and accurate timing also permits angiographic reconstructions of the three-dimensional images, producing MR angiography and MR venography that can be useful to better define the extent of a disease or for therapeutic planning.

3.2. Hepatocyte-targeted contrast agents 

Hepatocyte-targeted contrast agents are paramagnetic compounds that are partially taken up by the hepatocytes and excreted – to a variable extent – in the biliary tract. Two of these agents, gadobenate dimeglumine (Gd-BOPTA, MultiHance®, Bracco) and mangafodipir trisodium (MnDPDP, Teslascan®, Amersham Health) are currently available on the marked in Europe, whereas a third one, gadolinium ethoxybenzyl diethylenetriaminepentaacetic acid (Gd-EOB-DTPA, Schering AG) is undergoing phase III clinical trials. Gd-BOPTA and Gd-EOB-DTPA consist in a hydrophilic Gd-DTPA moiety covalently coupled to a lipophilic benzene ring. These compounds are amphiphilic and undergo both biliary and renal excretion. The hepatic uptake is thought to occur through the organic anion transport system located on the hepatocyte's membrane. In contrast, MnDPDP is a weak chelate of the manganese ion, and dissociates in vivo to give free manganese, which is taken up by the hepatocytes and excreted into the bile. The role of the ligand is to facilitate a slow release of the manganese, since high blood levels of free manganese may be toxic.

Hepatocyte-targeted contrast agents produce strong and sustained enhancement of liver parenchyma on T1-weighted images. Because of the selective increase in the signal intensity of normal hepatocytes compared with focal tumors, contrast-to-noise ratio between lesions and healthy liver is usually increased, and lesion detectability and conspicuity are improved on T1-weighted images [11], [12]. These agents have a very wide imaging window in the liver-specific phase, that ranges from 20 to 40 min to more than 2 h after the injection. Moreover, Gd-BOPTA and Gd-EOB-DTPA can be injected as a bolus, and can therefore be used – besides the liver-specific imaging – for serial dynamic scanning, like conventional gadolinium chelates [11], [13]. On the other hand, MnDPDP, owing to the higher biliary excretion rate, can also be used to perform contrast-enhanced MR cholangiography [14].

3.3. RES-targeted contrast agents 

RES-targeted contrast agents consist of suspensions of superparamagnetic iron oxide particles. Once injected intravenously, these particulate agents are rapidly removed from the circulation by the macrophage monocytic phagocytic system (or RES). Kupffer cells in the liver play a dominant role in this process, taking up more than 80% of circulating particles [15]. RES-targeted agents contain a magnetically responsive core (magnetite), which is made up of one or more aggregates of individual iron oxide crystals.

The effect of RES-targeted agents on signal intensity is twofold. A suspension of homogeneously dispersed superparamagnetic iron oxide particles is characterized by high T1 and T2 relaxivity. T1 relaxivity, however, requires intimate contact between water molecules and the surface of the iron oxides. As the T1 relaxivity is a function of the surface area, it increases when the particles are dispersed in the solvent and not concentrated in a small volume. Thus, the T1 effect (signal intensity increase on T1-weighted images) of these agents can be used during the vascular phase, before their RES uptake. In fact, as clustering of superparamagnetic iron oxide particles occurs because of the Kupffer cells fagocytosis, the surface of iron oxides in contact with water molecules decreases, and, therefore, the T1 relaxivity is greatly reduced. At this stage, the T2 effect (signal intensity decrease on T2-weighted images) due to dipole-dipole interactions and magnetic susceptibility becomes the dominant phenomenon. This effect is so potent, that even if the agents are concentrated in a small percentage of the liver volume (i.e. within the Kupffer cells), they can produce almost complete signal intensity loss on T2-weighted images in the entire normal liver.

The T2 effect of the late, liver-specific phase is the one commonly used for clinical applications. In fact, the reduction in signal intensity of normal liver parenchyma on T2-weighted images increases tumor-to-liver contrast and lesion detectability and conspicuity, since focal lesion – that usually do not contain Kupffer cells – do not change their signal intensity after injection of RES-targeted agents [15], [16]. Various types of superparamagnetic iron oxide particles with sizes ranging from 40 to 400 nm and with various coating materials have been investigated [15]. Two of these agents are currently available for clinical use in Europe: AMI-25 (Endorem®, Laboratoires Guerbet) and SH U 555 A (Resovist®, Schering AG). Peculiar features of SH U 555 A include bolus administration, that can therefore deliver dynamic imaging using the T1 effect of the vascular phase, and early Kupffer cells uptake, that allows liver-specific phase imaging to be performed as early as 10 min after the injection [17].

Back to Article Outline

4. Clinical application 

The clinical application of MRI in the evaluation of liver tumors is experiencing continuous growth. The diagnostic information provided by MRI and its current place in the imaging work-up of patients with known or suspected hepatic tumors are discussed in three main clinical scenarios: (a) characterization of incidental liver lesions; (b) detection and characterization of hepatocellular carcinoma (HCC) in cirrhosis; and (c) detection and characterization of liver metastases in oncology patients.

4.1. Characterization of incidental liver lesions 

Characterization of focal lesions of incidental detection is one of the most common and sometimes troublesome issues in liver imaging. Unsuspected lesions, in fact, are frequently detected in patients who have neither chronic liver disease nor history of malignancy during ultrasound examinations of the abdomen. While a confident diagnosis is usually made on the basis of ultrasound findings in cases of simple cysts and hemangiomas with typical hyperechoic appearance [18], lesions with non-specific ultrasound features require further investigation [19]. In this scenario, MRI is establishing a role as a primary diagnostic technique with respect to spiral CT. Spiral CT, in fact, has limitations in achieving a reliable diagnosis of small incidental lesions [19], [20]. In contrast, MRI – besides the information provided by the analysis of lesion signal intensity on baseline sequences and contrast enhancement patterns in dynamic studies – can offer improved capability in lesion characterization through the use of tissue-specific agents. In a recent multicenter comparative study of MRI with an hepatocyte-targeted contrast agent versus spiral CT, in which 145 patients who had or were suspected of having focal liver lesions were included, correct lesion classification was achieved by the blinded readers in 85% of cases with enhanced MRI and in only 68% with spiral CT [21].

MRI offers advantages over CT especially in the diagnosis of hemangioma and focal nodular hyperplasia (FNH). The accuracy of MRI in characterizing hemangiomas is in excess of 90% even with use of baseline sequences only [4]. The presence of slowly flowing blood within the vascular channels of the lesion gives to hemangiomas a consistent, marked hyperintensity on heavily T2-weighted images (Fig. 1). False diagnoses of hemangioma on baseline MRI studies are uncommon, but may occur with certain types of metastases, such as those of endocrine tumors, that may also show a very high signal intensity on T2-weighted images. When in doubt, a serial dynamic study after bolus injection of a gadolinium chelate can clarify the diagnosis. Dynamic MRI usually shows quite a typical perfusion pattern in hemangioma, with peripheral nodular enhancement in the early phase with centripetal progression to uniform or almost uniform enhancement during the portal venous and delayed phases (Fig. 1). Such a characteristic enhancement pattern has a specificity of more than 95% in diagnosing an hemangioma [22]. However, very small (less than 1.5 cm), high-flow hemangiomas frequently exhibit an hypervascular pattern, with uniform enhancement in the arterial phase, that may persist in the portal venous and delayed phases [22]. In these cases, diagnostic assessment may be difficult, and requires careful analysis of baseline and contrast-enhanced images. Hemangiomas show a peculiar feature after the injection of RES-targeted agents, that is lesion hyperintensity on T1-weighted post-contrast images. This behavior is due to the T1 effect of superparamagnetic iron oxide particles trapped within the slow-flow vascular channels of the lesion [15].

  • View full-size image.
  • Fig. 1. 

    Hepatic hemangioma. The lesion is hypointense on the T1-weighted image (a); and markedly hyperintense on the T2-weighted image (b). After administration of an extracellular fluid space contrast agent, peripheral nodular enhancement is detected in the arterial phase of the dynamic study (c), progressing centripetally to almost complete hyperintensity in the delayed phase (d).

MRI is also a very accurate method to characterize FNH. Owing to the affinity of its cells with normal hepatocytes, FNH is usually slighly hypointense or isointense with respect to normal liver parenchyma on T1-weighted images and slightly hyperintense or isointense on T2-weighted images [4], [23] (Fig. 2). The hallmark of the lesion, the central stellate scar, is clearly depicted because of its strong hypointensity on T1-weighted images and hyperintensity on T2-weighted images, reflecting its pathologic substratum of a vascularized connective tissue [4]. In large FNHs, diagnostic confirmation is usually achieved through a dynamic contrast-enhanced study [23], [24]. After bolus injection of a gadolinium chelate, FNH shows strong, early enhancement in the arterial phase, while it becomes isointense to liver parenchyma in the portal venous and delayed phases (Fig. 2). The central scar is usually unenhanced in the arterial phase, but may show contrast uptake in the delayed phase owing to the interstitial distribution of the paramagnetic agent. These lesion features on baseline and dynamic MRI have a specificity of more than 95% for the diagnosis of FNH [23], [24]. However, the central scar may not be detectable in small FNHs in both baseline and dynamic MRI studies [24], [25]. In these cases, the administration of a liver-specific agent may enable a more confident diagnosis. Owing to the affinity of its cells with the hepatocytes, FNH takes up hepatocyte-targeted agents, like normal parenchyma. These agents are then trapped within the lesion, since FNH is unable to effectively eliminate the compound via biliary excretion. Hence, it appears hyperintense to normal parenchyma on T1-weighted images. Also, the central scar – that does not take up the hepatocyte-targeted agent – becomes well delineated [24], [25] (Fig. 3). In one series, in which 100 FNHs were studied, correct characterization was achieved by administering a liver-specific hepatocyte-targeted agent in 90% of the FNHs with atypical features at the baseline and dynamic study [24]. The diagnosis of FNH has also been achieved with the use of RES-targeted agents. Because of its rich Kupffer cell population, FNH takes up iron oxide particles, and shows marked signal intensity decrease on T2-weighted images. The central scar is usually well delineated on post-contrast images as it does not contain RES cells and therefore keeps a high signal intensity [4], [25].

  • View full-size image.
  • Fig. 2. 

    Focal nodular hyperplasia. The lesion is isointense with respect to liver parenchyma on both the T1-weighted (a); and the T2-weighted image (b). The central scar is hypointense on the T1-weighted image (a); and hyperintense on the T2-weighted image (b). After administration of a gadolinium chelate, the lesion shows clear-cut enhancement in the arterial phase of the dynamic study, sparing the central scar (c). In the portal venous phase, the lesion becomes isointense to liver (d).

  • View full-size image.
  • Fig. 3. 

    Focal nodular hyperplasia. The small lesion, that was undetectable on baseline MRI, shows positive enhancement after administration of the hepatobiliary contrast agent MnDPDP. The central scar, that fails to take up the liver-specific agent, is well delineated.

While MRI allows a confident diagnosis of most hemangiomas and FNHs, characterization of other benign lesions is more problematic. Hepatocellular adenoma (HCA), although much less common than FNH, may arise in the same patient population. Unfortunately, MRI features of HCA are frequently non-specific [26]. The only feature that may suggest the diagnosis is the presence of intratumoral hemorrhage. MRI is in fact very sensitive in depicting areas of hemorrhage: they usually appear as markedly hyperintense regions on T1-weighted images, owing to the presence of extracellular metahemoglobine [4]. Even with the use of contrast agents, a final diagnosis can hardly be made in non-hemorrhagic HCAs. With dynamic MRI, HCA shows early enhancement during the arterial phase and becomes isointese to liver in the portal venous and delayed phases, features that overlap with those of other hypervascular tumors. On the other hand, the usefulness of tissue-specific MRI agents is not fully established. After administration of MnDPDP, HCA shows positive enhancement, but contrast uptake may not occur with Gd-BOPTA [26]. The variable Kupffer cell population of HCA does not gives to the lesion consistent features after the administration of RES-targeted agents.

4.2. Detection and characterization of HCC in cirrhosis 

Imaging evaluation of cirrhotic patients is a challenging issue. While the detection of a focal lesion in cirrhosis should always raise the suspicion of HCC, it is well established that the pathologic changes inherent in cirrhosis may simulate HCC in a variety of ways, especially because non-malignant hepatocellular lesions, such as regenerative and dysplastic nodules, may be indistinguishable from a small tumor. For years, the diagnosis of nodular lesions emerged in cirrhotic livers was based mainly on percutaneous biopsy. Currently, owing to the advances in imaging modalities, a reliable diagnostic assessment can be based in many instances on non-invasive examinations in combination with clinical and laboratory findings. Following the introduction of fast imaging techniques, MRI has become a tool for diagnosis and staging HCC.

HCC shows a variety of MRI features, that reflect the variable gross and microscopic characteristics of this malignancy. Tumor architecture, grading, stromal component, as well as intracellular content of certain substances, such as fat, glycogen, or metal ions greatly affect the MRI appearance of the lesion on T1-weighted and T2-weighted images [4]. The signal intensity may range from hypointensity to isointensity to hyperintensity on T1-weighted images and from isointensity to hyperintensity on T2-weighted images. Hyperintensity on T1-weighted images and isointensity on T2-weighted images are typical features of well-differentiated tumors, while hypointensity on T1-weighted images and hyperintensity on T2-weighted images are usually associated with moderately or poorly-differentiated tumors [4]. The signal intensity of HCC lesions may be inhomogeneous, reflecting the presence of areas with different degree of differentiation. Lesion signal intensity on baseline T1-weighted and T2-weighted images may help differentiate HCC from regenerative or dysplastic nodules in cirrhosis, but considerable overlap exists [4].

One of the key pathologic factors for differential diagnosis between HCC and regenerative or dysplastic nodules that is reflected in imaging appearances is the vascular supply to the lesion. Through the progression from regenerative nodule, to low-grade dysplastic nodule, to high-grade dysplastic nodule, to frank HCC, one sees loss of visualization of portal tracts and development of new arterial vessels, termed non-triadal arteries, which become the dominant blood supply in overt HCC lesions. It is this neovascularity that allows HCC to be diagnosed and is the key for imaging cirrhotic patients [27]. Current dynamic contrast-enhanced MRI scanning protocols allow selective imaging of the entire liver in the arterial, the portal venous, and the delayed phases, such as in spiral CT protocols [28], [29], [30]. Dynamic MRI well demonstrates the hallmark of HCC in the cirrhotic liver, that is, arterial phase enhancement with portal venous phase wash-out (Fig. 4). This feature enables one to distinguish frank HCC from dysplastic nodules, that usually are not hypervascular on arterial phase MRI, and enhance homogeneously on portal venous phase imaging, appearing isointense or nearly isointense to surrounding liver tissue.

  • View full-size image.
  • Fig. 4. 

    Hepatocellular carcinoma. The small lesion is undetectable on the T1-weighted image (a). After administration of an extracellular fluid space contrast agent, the nodule shows clear-cut enhancement in the arterial phase of the dynamic study (b), with rapid wash-out in the portal venous phase (c).

While the dynamic study performed by using gadolinium chelates is a currently a key part of the MRI examination, liver-specific contrast agents have also been used in attempts to improve the information provided by MRI in HCC detection and characterization. It has been shown that some, well-differentiated HCCs may show a positive enhancement after the administration of MnDPDP because of their affinity with normal hepatocytes. In one study, owing to this peculiar feature, early-stage tumors that were missed by spiral CT because of their immature neovascularity were detected [31]. However, since uptake of MnDPDP also occurs in dysplastic nodules, differential diagnosis among these entities can not be achieved [31]. HCC conspicuity after the administration of RES-specific contrast agents depends on differences in the number of Kupffer cells within the nodule and the surrounding cirrhotic liver [32]. While moderately or poorly differentiated HCCs containing few or no Kupffer cells show high contrast-to-noise ratio (Fig. 5), well-differentiated HCCs (as well as dysplastic nodules) have a Kupffer cell population that may not significantly differ from that of surrounding parenchyma, which results in a signal-to-noise ratio close to zero and, thus, in low detectability rates [32]. Despite in one series MRI with use of a RES-targeted agent was superior to spiral CT for the detection of HCC nodules [33], in a comparative study the sensitivity of iron oxide-enhanced imaging in detection of small HCCs was inferior to that of gadolinium-enhanced MRI [34]. In addition, the specificity may not improve after the administration of RES-targeted agents because of the false-positive lesions that may be caused by fibrotic changes [35].

  • View full-size image.
  • Fig. 5. 

    Hepatocellular carcinoma. Baseline T1-weighted image hardly shows slightly hypointense tumor in segment 3 (a). In the arterial phase image of the dynamic study performed after bolus injection of the superparamagnetic iron oxide agent SH U 555 A, tumor is hyperintense because of its hypervascularity (b). T2-weighted image in the liver-specific phase of the RES-targeted agent clearly delineates the lesion owing to its hyperintensity compared with the blackened liver (c).

The place of MRI in the diagnostic work-up of nodular lesions in cirrhosis has been defined [36]. The diagnostic algorithm takes into account the actual risk of malignancy according to the size of the lesion and the possibility to achieve a reliable diagnosis [36]. In nodules smaller than 1 cm detected during ultrasound surveillance in cirrhotic patients, in view of the high prevalence of non-malignant lesions and the difficulties in achieving a final diagnosis of HCC, a reasonable protocol is to repeat ultrasound every 3 months. When the nodule exceeds 1 cm in size, diagnostic confirmation should be pursued. If the nodule is in the range of 1–2 cm, biopsy may be still recommended, since imaging techniques do not seem to have sufficient accuracy to distinguish HCC from benign conditions [36]. In one study, 54 of 104 (52%) of small (less than 2 cm), round or oval, early-enhancing hepatic lesions at serial contrast-enhanced dynamic MRI were not confirmed as HCC at follow-up [37]. For nodules above 2 cm, it is accepted that imaging techniques may confidently establish the diagnosis without needing confirmation with a positive biopsy. However, coincident findings from at least two imaging studies are required for non-invasive diagnosis [36]. Thus, MRI – with use of baseline plus serial gadolinium-enhanced dynamic imaging – may be used as an alternate to spiral CT to integrate ultrasound findings.

Nevertheless, despite technical improvements, MRI remains relatively insensitive for the detection of tiny satellite nodules associated with the main tumor and for the identification of tumor vascular invasion into peripheral portal vein branches. When careful lesion-by-lesion imaging-pathologic correlation was performed, the results of MRI in the diagnosis of small HCC nodules were shown to be much worse than previously estimated [38], [39], [40], [41], [42], [43]. In series in which the results of MRI were correlated with histopathologic results after thin-section slicing of the explanted liver, lesion-by-lesion analysis revealed a sensitivity of only 53–78%, with positive predictive values ranging 56–89% (Table 2). Nevertheless, state-of-the-art dynamic contrast-enhanced MRI – performed by using angiographic three-dimensional sequences – was shown to provide higher detection rates than spiral CT, especially in lesions in the range of 1–2 cm [41]. Therefore, MRI is currently considered in several centers as the optimal technique for accurate staging of HCC patients who are candidates to radical therapies [40], [41].

Table 2. Sensitivity and positive predictive value of gadolinium-enhanced dynamic magnetic resonance imaging (MRI) and spiral computed tomography (CT) in the diagnosis of hepatocellular carcinoma (HCC) according to lesion size
AuthorNo. of patients/no. of lesionsOveralllesion sensitivity (%)Sensitivity for lesions <1 cm (%)Sensitivity for lesions 1–2 cm (%)Sensitivity for lesions >2 cm (%)Positive predictive value (%)
Krinsky et al. [38]71/19MRI, 10/19 (53)1/3 (33)6/12 (50)3/4 (75)19/34 (56)
CT, N/P

Rode et al. [39]

43/13MRI, 10/13 (77)MRI, 5/7 (71)MRI, 4/5 (80)MRI, 1/1 (100)MRI, N/A
CT, 7/13 (54)CT, 3/7 (43)CT, 3/5 (60)CT, 1/1 (100)CT, N/A

de Lédinghen

34/54MRI, 33/54 (61)MRI, 2/8 (25)MRI, 19/34 (56)MRI, 12/12 (100)MRI, 33/37 (89)
et al. [40] CT, 28/54 (52)CT, 2/8 (25)CT, 15/34 (44)CT, 11/12 (92)CT, 28/37 (76)

Llovet et al. [41]

50/76MRI, 56/76 (74)MRI, 7/22 (32)aMRI, 16/19 (84)aMRI, 6/6 (100)aN/A
CT, N/ACT, N/ACT, 9/19 (47)CT, N/A

Bhartia et al. [42]

31/32MRI, 25/32 (78)b3/8 (38)12/13 (92)10/11 (91)25/46 (54)
CT, N/P

Teefey et al. [43]

22/18MRI, 14/18 (77)N/AN/AN/AMRI, 14/19 (74)
CT, 13/18 (72) CT, 13/22 (59)

All series have pathologic examination of the explanted liver as term of reference. Note. N/P, not performed; N/A, not available.

a Analysis of sensitivity according to lesion size was focused on 47 satellite nodules and did not include 29 main tumors.

b MRI examination protocol included baseline, dynamic gadolinium-enhanced, and ferumoxide-enhanced imaging.

4.3. Detection and characterization of metastases in oncology patients 

Metastatic disease involving the liver is one of the most common issues in oncology. Spiral CT is the standard imaging technique in most oncology protocols. It has the advantage to provide objective documentation of the extent of the liver tumor burden, which may be useful to monitor tumor response over time, and to effectively assess extrahepatic disease. Nevertheless, MRI may be a tool in this scenario for selected clinical applications. A well-established indication for MRI is a questionable liver metastasis identified by previous ultrasound or spiral CT, provided that a precise information is important to define patient's prognosis or for treatment choice. Typically, this diagnostic issue concerns small focal lesions, in which the features at contrast-enhanced spiral CT may not enable a confident differentiation between a metastasis or a benign lesion, such as an hemangioma or an area of focal fatty sparing [24].

Moreover, MRI is establishing a role in selected oncology patients for whom exact knowledge of the number and location of metastatic lesions is crucial for treatment planning. This is the case of patients with limited hepatic metastatic disease from certain primary malignancies of the gastrointestinal tract, especially colorectal adenocarcinoma, that are possible candidates for surgical resection or percutaneous tumor ablation. Over the years, several CT and MRI examination protocols have been used for preoperative detection of colorectal liver metastases. Until recently, the general opinion was that CT arterial portography was the best preoperative imaging method, despite its invasiveness and high false-positive rate. The sensitivity of baseline MRI, in fact, was equal or at best only marginally higher than that of contrast-enhanced CT [44]. Also, the use of dynamic MRI protocols with extracellular fluid space contrast agents did not lead to any substantial improvement in lesion detectability, because of the hypovascular nature of colorectal liver metastases [45].

However, the advent of tissue-specific agents has greatly improved the sensitivity of MRI in detecting liver metastases. Because metastases do not exhibit hepatocyte-like features and do not contain Kupffer's cells, no uptake of hepatocyte-selective or RES-targeted agents is seen within the lesions. As a result, selective enhancement of normal hepatic parenchyma can be achieved with either tissue-specific compounds, resulting in substantial increase in tumor-to-liver contrast and therefore in improved lesion detectability and conspicuity (Fig. 6). Several studies have shown that MRI with use of hepatocyte-selective or RES-targeted agents is superior to any spiral CT protocols for pretreatment assessment of colorectal metastases [47], [48], [49], [50], [51]. In series of hepatic colorectal cancer metastatic patients in which careful lesion-by-lesion correlation between imaging and intraoperative or pathology findings was performed, the sensitivity of liver-specific MRI ranged 68–90% compared to 49–74% of spiral CT (Table 3).

  • View full-size image.
  • Fig. 6. 

    Colorectal liver metastases. Baseline T1-weighted image shows solitary lesion in segment 7 (a). After administration of the hepatobiliary contrast agent MnDPDP, lesion conspicuity is increased, and an additional tiny lesion, undetectable on baseline scans, is detected in segment 2 (b).

Table 3. Comparison of contrast-enhanced magnetic resonance imaging (MRI) and spiral computed tomography (CT) in the detection of liver metastases in series with intraoperative ultrasound or pathologic correlations
AuthorNo. of patients/no. of lesionsMRI protocolDetection rate of spiral CTDetection rate of MRI
Strotzer et al. [46]35/53Baseline+SPIO-enhanced26/53 (49%)38/53 (72%)a
Lencioni et al. [47]14/36Baseline+SPIO-enhanced21/36 (58%)30/36 (83%)a
Ward et al. [48]31/81Baseline+SPIO-enhanced74%b81%a,b
Bluemke et al. [49]32/71Baseline+SPIO-enhanced60%b68%a,b
Mann et al. [50]20/64Baseline+MnDPDP-enhanced61%83%
Bartolozzi et al. [51]44/128Baseline+MnDPDP-enhanced91/128 (71%)115/128 (90%)a

Note. N.A., Not available. SPIO, superparamagnetic iron oxide particle. MnDPDP, mangafodipir trisodium.

a Detection rate of MRI was significant higher than that of spiral CT at statistical analysis.

b Mean detection rate of blinded observers.

MRI with tissue-specific contrast agents currently stands as the most accurate imaging technique for individual hepatic metastasis detection and characterization even with respect fluoro-18-deoxyglucose positron emission tomography [52]. While the initial diagnostic approach to oncology patients will vary by tumor histotype as well as by institution, a liver-specific MRI examination should be offered for pretreatment assessment of patients in whom metastases are of a pathologic type for which surgical resection or local tumor ablation have proved effective and less expensive routine screening examinations have depicted only a number of metastases that makes surgery or tumor ablation a feasible option.

Back to Article Outline

References 

  1. Keogan MT, Edelman RR. Technologic advances in abdominal MR imaging. Radiology. 2001;220:310–320
  2. Hussain HK, Londy FJ, Francis IR, Nghiem HV, Weadock WJ, Gebremariam A, et al.  Hepatic arterial phase MR imaging with automated bolus-detection three-dimensional fast gradient-recalled-echo sequence: comparison with test-bolus method. Radiology. 2003;226:558–566
  3. Semelka RC, Helmberger KG. Contrast agents for MR imaging of the liver. Radiology. 2001;218:27–38
  4. Bartolozzi C, Cioni D, Donati F, Lencioni R. Focal liver lesions: MR imaging-pathologic correlation. Eur Radiol. 2001;11:1374–1388
  5. Semelka RC, Balci NC, Op de Beeck B, Reinhold C. Evaluation of a 10-min comprehensive MR imaging examination of the upper abdomen. Radiology. 1999;211:189–195
  6. Augui J, Vignaux O, Argaud C, Coste J, Gouya H, Legmann P. Liver: T2-weighted MR imaging with breath-hold fast-recovery optimized fast spin-echo compared with breath-hold half-Fourier and non-breath-hold respiratory-triggered fast spin-echo pulse sequences. Radiology. 2002;223:853–859
  7. Laissy JP, Trillaud H, Douek P. MR angiography: non-invasive vascular imaging of the abdomen. Abdom Imaging. 2002;27:488–506
  8. Nelson KL, Gifford LM, Lauber-Huber C, Gross CA, Lasser TA. Clinical safety of gadopentetate dimeglumine. Radiology. 1995;196:439–443
  9. Lencioni R, Mascalchi M, Paolicchi A, Zampa V. Breath-hold spin-echo MR imaging for evaluation of dynamic enhancement of native and treated hepatocellular carcinoma after intravenous Gd-DTPA administration. MAGMA. 1995;3:151–156
  10. Yu JS, Kim KW, Kim EK, Lee JT, Yoo HS. Contrast enhancement of small hepatocellular carcinoma: usefulness of three successive early image acquisitions during multiphase dynamic MR imaging. AJR Am J Roentgenol. 1999;173:597–604
  11. Petersein J, Spinazzi A, Giovagnoni A, Soyer P, Terrier F, Lencioni R, et al.  Focal liver lesions: evaluation of the efficacy of gadobenate dimeglumine in MR imaging – a multicenter phase III clinical study. Radiology. 2000;215:727–736
  12. Braga HJ, Choti MA, Lee VS, Paulson EK, Siegelman ES, Bluemke DA. Liver lesions: manganese-enhanced MR and dual-phase helical CT for preoperative detection and characterization comparison with receiver operating characteristic analysis. Radiology. 2002;223:525–531
  13. Reimer P, Rummeny EJ, Daldrup HE, Hesse T, Balzer T, Tombach B, et al.  Enhancement characteristics of liver metastases, hepatocellular carcinomas, and hemangiomas with Gd-EOB-DTPA: preliminary results with dynamic MR imaging. Eur Radiol. 1997;7:257–280
  14. Kapoor V, Peterson MS, Baron RL, Patel S, Eghtsesad B, Fing JJ. Intrahepatic biliary anatomy of living adult liver donors: correlation of mangafodipir trisodium-enhanced MR cholangiography and intraoperative cholangiography. AJR Am J Roentgenol. 2002;179:1281–1286
  15. Reimer P, Tombach B. Hepatic MRI with SPIO: detection and characterization of focal liver lesions. Eur Radiol. 1998;8:1198–1204
  16. Ros PR, Freeny PC, Harms SE, Seltzer SE, Davis PL, Chan TW, et al.  Hepatic MRI imaging with ferumoxide: a multicentric clinical trial of the safety and efficacy in the detection of focal hepatic lesions. Radiology. 1995;196:481–488
  17. Kopp AF, Laniado M, Dammamm F, Stern W, Gronewaller E, Balzer T, et al.  MR imaging of the liver with Resovist: safety, efficacy, and pharmacodynamic properties. Radiology. 1997;204:749–756
  18. Leifer DM, Middleton WD, Teefey SA, Menias CO, Leahy JR. Follow-up of patients at low risk for hepatic malignancy with a characteristic hemangioma at US. Radiology. 2000;214:167–172
  19. Vilgrain V, Boulos L, Vullierme MP, Denys A, Terris B, Menu Y. Imaging of atypical hemangiomas of the liver with pathologic correlation. RadioGraphics. 2000;20:379–397
  20. Kim T, Federle MP, Baron RL, Peterson MS, Kawamori Y. Discrimination of small hepatic hemangiomas from hypervascular malignant tumors smaller than 3 cm with three-phase helical CT. Radiology. 2001;219:699–706
  21. Oudkerk M, Torres CG, Song B, Konig M, Grimm J, Fernandez-Cuadrado J, et al.  Characterization of liver lesions with mangafodipir trisodium-enhanced MR imaging: multicenter study comparing MR and dual-phase spiral CT. Radiology. 2002;223:517–524
  22. Semelka RC, Brown ED, Ascher SM, Patt RH, Bagley AS, Li W, et al.  Hepatic hemangiomas: a multi-institutional study of appearance on T2-weighted and serial gadolinium-enhanced gradient-echo MR images. Radiology. 1994;192:401–406
  23. Mortele KJ, Praet M, Van Vlierberghe H, de Hemptinne B, Zou K, Ros PR. Focal nodular hyperplasia of the liver: detection and characterization with plain and dynamic-enhanced MRI. Abdom Imaging. 2002;27:700–707
  24. Grazioli L, Morana G, Federle MP, Brancatelli G, Testoni M, Kirchin MA, et al.  Focal nodular hyperplasia: morphologic and functional information from MR imaging with gadobenate dimeglumine. Radiology. 2001;221:731–739
  25. Ba-Ssalamah A, Schima W, Schmook MT, Linnau KF, Schibany N, Helbich T, et al.  Atypical focal nodular hyperplasia of the liver: imaging features of non-specific and liver-specific MR contrast agents. AJR Am J Roentgenol. 2002;179:1447–1456
  26. Grazioli L, Federle MP, Brancatelli G, Ichikawa T, Olivetti L, Blachar A. Hepatic adenomas: imaging and pathologic findings. RadioGraphics. 2001;21:877–892
  27. Lencioni R, Mascalchi M, Caramella D, Bartolozzi C. Small hepatocellular carcinoma: differentiation from adenomatous hyperplasia with color Doppler US and dynamic Gd-DTPA-enhanced MR imaging. Abdom Imaging. 1996;21:41–48
  28. Noguchi Y, Murakami T, Kim T, Hori M, Osuga K, Kawata S, et al.  Detection of hypervascular hepatocellular carcinoma by dynamic magnetic resonance imaging with double-echo chemical shift in-phase and opposed-phase gradient echo technique: comparison with dynamic helical computed tomography imaging with double arterial phase. J Comput Assist Tomogr. 2002;26:981–987
  29. Eubank WB, Wherry KL, Maki JH, Sahin H, Funkhouser CP, Schmiedl UP. Preoperative evaluation of patients awaiting liver transplantation: comparison of multiphasic contrast-enhanced 3D magnetic resonance to helical computed tomography examinations. J Magn Reson Imaging. 2002;16:565–575
  30. Noguchi Y, Murakami T, Kim T, Hori M, Osuga K, Kawata S, et al.  Detection of hepatocellular carcinoma: comparison of dynamic MR imaging with dynamic double arterial phase helical CT. AJR Am J Roentgenol. 2003;180:455–460
  31. Bartolozzi C, Donati F, Cioni D, Crocetti L, Lencioni R. MnDPDP-enhanced MRI vs dual-phase spiral CT in the detection of hepatocellular carcinoma in cirrhosis. Eur Radiol. 2000;10:1697–1702
  32. Lim JH, Choi D, Cho SK, Kim SH, Lee WJ, Lim HK, et al.  Conspicuity of hepatocellular nodular lesions in cirrhotic livers at ferumoxides-enhanced MR imaging: importance of Kupffer cell number. Radiology. 2001;220:669–676
  33. Kang BK, Lim JH, Kim SH, Choi D, Lim HK, Lee WJ, et al.  Preoperative depiction of hepatocellular carcinoma: ferumoxides-enhanced MR imaging versus triple-phase helical CT. Radiology. 2003;226:79–85
  34. Pauleit D, Textor J, Bachmann R, Conrad R, Flacke S, Layer G, et al.  Hepatocellular carcinoma: detection with gadolinium- and ferumoxides-enhanced MR imaging of the liver. Radiology. 2002;222:73–80
  35. Mori K, Scheidler J, Helmberger T, Holzknecht N, Schauer R, Schirren CA, et al.  Detection of malignant hepatic lesions before orthotopic liver transplantation: accuracy of ferumoxides-enhanced MR imaging. AJR Am J Roentgenol. 2002;179:1045–1051
  36. Bruix J, Sherman M, Llovet JM, Beaugrand M, Lencioni R, Burroughs AK, et al.  EASL Panel of Experts on HCC. Clinical management of hepatocellular carcinoma. Conclusions of the Barcelona-2000 EASL conference. European Association for the Study of the Liver. J Hepatol. 2001;35:421–430
  37. Shimizu A, Ito K, Koike S, Fujita T, Shimizu K, Matsunaga N. Cirrhosis or chronic hepatitis: evaluation of small (<or=2-cm) early-enhancing hepatic lesions with serial contrast-enhanced dynamic MR imaging. Radiology. 2003;226:550–555
  38. Krinsky GA, Lee VS, Theise ND, Weinreb JC, Rofsky NM, Diflo T, et al.  Hepatocellular carcinoma and dysplastic nodules in patients with cirrhosis: prospective diagnosis with MR imaging and explantation correlation. Radiology. 2001;219:445–454
  39. Rode A, Bancel B, Douek P, Chevallier M, Vilgrain V, Picaud G, et al.  Small nodule detection in cirrhotic livers: evaluation with US, spiral CT, and MRI and correlation with pathologic examination of explanted liver. J Comput Assist Tomogr. 2001;25:327–336
  40. de Lédinghen V, Laharie D, Lecesne R, Le Bail B, Winnock M, Bernard PH, et al.  Detection of nodules in liver cirrhosis: spiral computed tomography or magnetic resonance imaging? A prospective study of 88 nodules in 34 patients. Eur J Gastroenterol Hepatol. 2002;14:159–165
  41. Burrel M, Llovet JM, Ayuso C, Iglesias C, Sala M, Miquel R, et al.  MRI angiography is superior to helical CT for HCC detection prior to liver transplantation: an explant correlation. Hepatology. 2003;38:1034–1042
  42. Barthia B, Ward J, Guthrie JA, Robinson PJ. Hepatocellular carcinoma in cirrhotic livers: double-contrast thin-section MR imaging with pathologic correlation of explanted tissue. AJR Am J Roentgenol. 2003;180:577–584
  43. Teefey SA, Hildeboldt CC, Dehdashti F, Siegel BA, Peters MG, Heiken JP, et al.  Detection of primary hepatic malignancy in liver transplant candidates: prospective comparison of CT, MR imaging, US, and PET. Radiology. 2003;226:533–542
  44. Bartolozzi C, Lencioni R, Donati F, Cioni D. Abdominal MR: liver and pancreas. Eur Radiol. 1999;9:1496–1512
  45. Hamm B, Mahfouz AE, Taupitz M, Mitchell DG, Nelson R, Halpern E, et al.  Liver metastases: improved detection with dynamic gadolinium-enhanced MR imaging?. Radiology. 1997;202:677–682
  46. Strotzer M, Gmeinwieser J, Schmidt J, Fellner C, Seitz J, Albrich H, et al.  Diagnosis of liver metastases from colorectal adenocarcinoma. Acta Radiol. 1997;38:986–992
  47. Lencioni R, Donati F, Cioni D, Paolicchi A, Cicorelli A, Bartolozzi C. Detection of colorectal liver metastases: prospective comparison of unenhanced and ferumoxides-enhanced magnetic resonance imaging at 1.5 T, dual-hase spiral CT, and spiral CT during arterial portography. MAGMA. 1998;7:76–87
  48. Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ. Hepatic lesion detection: comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative-free response receiver operating characteristic analysis. Radiology. 1999;210:459–466
  49. Bluemke DA, Paulson EK, Choti MA, DeSena S, Clavien PA. Detection of hepatic lesions in candidates for surgery: comparison of ferumoxides-enhanced MR imaging and dual-phase helical CT. AJR Am J Roentgenol. 2000;175:1653–1658
  50. Mann GN, Marx HF, Lai LL, Wagman LD. Clinical and cost effectiveness of a new hepatocellular MRI contrast agent, mangafodipir trisodium, in the preoperative assessment of liver resectability. Ann Surg Oncol. 2001;8:573–579
  51. Bartolozzi C, Donati F, Cioni D, Procacci C, Morana G, Chiesa A, et al.  Detection of colorectal liver metastases: a prospective multicenter trial comparing unenhanced MRI, MnDPDP-enhanced MRI, and spiral CT. Eur Radiol. 2003; in press
  52. Yang M, Martin DR, Karabulut N, Frick MP. Comparison of MR and PET imaging for the evaluation of liver metastases. Magn Reson Imaging. 2003;17:343–349

PII: S0168-8278(03)00455-0

doi:10.1016/S0168-8278(03)00455-0

Journal of Hepatology
Volume 40, Issue 1 , Pages 162-171, January 2004

Article Tools

* Image Usage & Resolution