The last 10 years have seen dramatic changes in the therapeutic approach to primary and secondary liver malignancies. This has been due, in part, to the increasing proliferation of new less-invasive treatments, including percutaneous ablative therapies (ethanol injection, radio-frequency ablation, laser ablation), angiographic therapies (segmental chemoembolization), and liver transplantation, in addition to wider applications of pre-existing treatments (resection and systemic chemotherapy) either alone or in conjunction with the new treatments. The broad array of treatments available requires current medical management to select the therapeutic approach for each patient, which will provide the highest success rate while taking into consideration the risks, costs, and availability of the required technology.

Recent advances in ultrasound (US), computed tomography (CT), and magnetic resonance (MR) imaging mean that pretreatment imaging of the liver is now the critical step in evaluating patients with liver malignancies. The imaging of patients with suspected liver masses can be viewed as having three purposes: detection, characterization, and the assessment of intrahepatic and extrahepatic extent of tumor. Given the high incidence of benign liver tumors, ^1,2 it is fairly common to detect these lesions both incidentally and during the work-up of oncologic patients. Thus, while lesion detection is a principal objective of liver imaging, lesion characterization is also an important concurrent objective. Similarly, the accurate staging of disease extent is a prerequisite for the selection of appropriate therapy.

The advances in diagnostic imaging have enhanced our ability to detect and characterize focal liver lesions, affording a definitive diagnosis noninvasively now more than ever before. The information acquired, combined with appropriate clinical information, facilitates the preparation of optimal treatment plans for patients with liver malignancies. This review will summarize some of the recent advances in MR imaging of liver masses, focusing on contrast agents and the relative role of MR to CT and US in evaluating patients with liver masses.

LIVER LESION DETECTION: THE MEDICAL NEED

Liver Lesion Detection in the Noncirrhotic Population

This is commonly undertaken in two distinct clinical situations: in patients with known malignancy for whom staging of disease is required, and in patients with known liver malignancy who are considered candidates for local resection or interventional treatment. Patients with a known extrahepatic malignancy require evaluation for abdominal metastases for staging extent of disease. In this setting, the imaging modality of choice should be one that most effectively and efficaciously can simultaneously evaluate the liver and the extrahepatic abdomen. Contrast-enhanced CT is generally accepted as the most appropriate modality for such a screening study, primarily due to the ease of performing and interpreting large numbers of such examinations, its widespread availability, and a generally acknowledged superior ability to evaluate the extrahepatic abdomen. ³ In addition, if other body parts, such as the head or chest, need to be evaluated for the spread of disease, they can all be examined in a single visit according to a cost-effective, efficient protocol (Figure 1).

Although MR currently plays only a limited role in this setting, it is appropriate in certain specific clinical situations. While not as easily utilized for screening large numbers of patients on a daily basis, MR can certainly evaluate the abdomen efficiently and can be utilized in patients who are allergic to iodinated contrast media. A clinical situation in which MRI is the preferred liver imaging modality is in the presence of fatty infiltration of the liver in patients with potential liver malignancies. Fatty infiltration is a common problem that often results from poor nutrition, the effects of chemotherapy, or regional differences in liver perfusion in patients with malignancy. In these situations, diffuse fatty changes can often obscure the presence of underlying focal lesions on CT (Figure 2). Moreoever, focal fatty infiltration or sparing can often create pseudolesions on CT scans. ⁴ MR, especially with opposed-phase gradient-echo techniques, is superior to CT for detecting or confirming the presence of lesions against a background of fatty liver (Figure 3).

The second important clinical setting in which lesion detection is vital is in patients with a known liver malignancy who are being evaluated for liver resection or lesion ablation. This is particularly relevant in patients with colorectal cancer, where up to 30% of patients may have isolated liver metastases. It is in this sub-group that local therapy, in particular liver resection, can offer the possibility of cure. Studies have shown a 5-year survival rate of 20% to 40% following partial hepatic resection of colorectal metastases in selected patients, compared with a 5-year survival rate of 5% to 10% in untreated patients. ^5-8 However, successful outcome depends greatly on an accurate knowledge of the size and location of the tumor burden. Imaging plays a vital role in differentiating patients with limited local tumor disease, who may benefit from radical surgery or interventional treatments, from patients with widespread disease for whom such treatments are unlikely to be beneficial.

While multidetector helical CT has greatly improved the ability of CT to evaluate the liver in patients with malignancy, ⁹ studies with strong pathologic correlation have yet to demonstrate adequate capacity for the detection of small (<1.5 cm) tumor nodules. ¹⁰ Until recently, CT during arterial portography (CTAP) was widely used for the preoperative staging of liver disease with a sensitivity of approximately 90% for lesion detection, compared with 70% for conventional contrast-enhanced CT (CECT). ^11-13 The advent of multidetector helical CT has significantly improved the sensitivity of CT to approaching 90% for dynamic helical CT, and to >90% for helical CTAP. ^14-16 However, CTAP is an invasive technique and has low specificity due to numerous benign perfusion defects. ¹⁷ Moreover, concern over the nephrotoxicity of iodinated contrast agents and the requisite use of ionizing radiation in all CT examinations are factors to be considered when referring patients for diagnostic evaluation of the liver.

Contrast-enhanced MR provides a viable alternative to helical CT for all liver examinations. Traditionally, contrast-enhanced liver MR has been performed with conventional gadolinium chelates, which distribute exclusively to the extracellular space and rely on differences in perfusion between normal liver and lesion to improve the detection of liver lesions beyond that achievable on unenhanced MR alone. ^18,19 However, while such agents have proven beneficial for the improved detection of hypervascular tumors, they have not always been shown to increase the detection of hypovascular metastases. ^20,21 The development and use of contrast-enhanced MRI using contrast agents with liver-specific properties is emerging as an appropriate modality for the preoperative staging of liver malignancy; in most practices it has replaced CTAP (Figure 4).

Liver Lesion Detection in the Cirrhotic Population

There is a very high incidence of hepatocellular carcinoma (HCC) in patients with chronic liver disease, particularly hepatitis B and C, long-standing cirrhosis, and hemochromatosis. Studies of patients referred for transplantation have consistently demonstrated >10% incidence of HCC in patients ultimately transplanted for severe cirrhosis, and approximately a 20% incidence in patients with hepatitis B and C. ^22,23 Despite the use of US and serum alpha-fetoprotein determinations as screening techniques for the detection of HCC in cirrhotic patients, both are characterized by low sensitivity when reported in screening population studies with pathologic correlation. ^24-26 The detection rate of HCC with US is related to the size, location, and echotexture of the lesions, as well as to the US technology employed and the experience of the operator. While US HCC detection rates from 82% to 93% have been reported for tumors measuring between 2 and 3 cm, ^27-29 results at these high sensitivities have not been confirmed in screening populations or in studies with pathologic confirmation. The reported rate for HCCs <2 cm in diameter is considerably more variable and is generally accepted as inadequate for tumor detection. The detection rate on US for HCCs <1 cm is even lower, ranging between 13% and 37%. ^28,30 Generally, detection, especially of small nodules, is severely affected by the location and echogenicity of the tumor and by the level of expertise on the part of the operator. Additionally, diffuse HCCs are very difficult to identify and characterize with US, since the diffuse parenchymal inhomogeneity can be misinterpreted as being due to cirrhosis (Figure 5). The use of ultrasound is further complicated by the numerous false-positive lesions created by the nodular regeneration inherent in cirrhosis. ³¹

The sensitivity of US for the detection of HCCs can be reasonably increased by improving contrast resolution, ie, by optimizing US beam focalization and frequency, preferably with the most updated wideband transducers, and/or by the use of tissue harmonic imaging. The recent advent of contrast agents for ultrasound (eg, Levovist, Schering AG, Berlin, Germany; SonoVue, Bracco Imaging SpA, Milan, Italy) may further improve the impact of US for liver lesion detection, ^32-34 but the clinical utility of these remain to be proven in large populations and will need to be compared with CT and MR.

Evaluating the cirrhotic liver for HCC is further confounded by the multiple causes of nodularity in these patients. First and foremost, HCC is a multifocal tumor and daughter nodules or multiple sites of tumor development are common features. All imaging modalities tend to underestimate the number of multicentric nodules and daughter HCCs in these patients. This is particularly evident in studies in which end-stage resected cirrhotic liver with pathologic examination following transplantation are compared with previous radiologic findings. A study of 46 patients undergoing orthotopic liver transplantation with pathologic examination of the explanted livers, revealed pretransplantation sensitivities of 58% for CT with iodized oil, 67% for digital subtraction angiography and 85% for CTAP. Most false-negative diagnoses in this study were due to either small HCCs (<2 cm) or intrahepatic metastases. ³⁵

The sensitivity for detecting HCCs <3 cm in diameter has a wide variability in literature reports. This wide range of detection rates probably reflects the population being studied more than the abilities of a modality or the investigator. Screening studies with pathologic proof in patients without suspected tumor have consistently shown that the highest detection rate for HCC detection in cirrhosis for any modality is not accurate. The diagnostic modality with the highest reported success rate is intraoperative US, with sensitivities ranging from 90% to 98%. ^36-40 With intraoperative US, the issue of false-positive diagnoses is less often of importance, as the issue can be addressed directly.

Currently, biphasic or triphasic helical CT achieves a higher detection rate than does conventional CT, with the arterial phase proving more accurate than the portal-venous phase, especially for the detection of HCCs <2 cm in size. ^10,41,42

The sensitivity achieved by current MRI technology is essentially equal to that of helical CT, especially if conventional T1- and T2-weighted images with various sequences are coupled with gadolinium chelate dynamic contrast-enhanced images (Figure 6). While multidetector CT can obtain two arterial-phase passes (early and late), which may improve HCC detection, ^9,43-45 MRI has the ability to obtain multiple arterial passes and has the added advantage of doing this without additional radiation. This can improve HCC detection. ^20,46 In addition, contrast agents with hepatocyte-specific properties have been shown to improve the detection of HCC. ^41,47,48

LIVER LESION CHARACTERIZATION

The liver is a common repository for many benign and malignant focal lesions. Many of the benign entities, such as hemangioma, focal nodular hyperplasia, cysts, and focal fatty change are very common and the importance for radiologists is to recognize them as such, since these lesions usually require no further attention. In contrast, malignant lesions in the liver may require prompt surgical or interventional attention. The primary malignancies that occur in the liver may arise from hepatocytes, bile duct epithelium, endothelial cells, or lymphoid cells. The majority of malignancies encountered in clinical practice are epithelial in origin, with HCC and cholangiocarcinoma accounting for nearly all. Whereas HCC is the most common abdominal primary malignancy worldwide, the liver is also a very fertile soil for metastases because of: 1) its dual blood supply from the systemic and splanchnic systems; 2) the presence of humoral factors that promote cell growth; and 3) the discontinuous nature of the endothelial lining of the hepatic sinusoid, which allows open communication with the extracellular space of Disse. ⁴⁹

The predominance of benign lesions in the liver, even in patients with known malignancies elsewhere, is clear from a recent retrospective study reporting 2978 cancer patients who had undergone CT examinations. ² Small (<1 cm) hepatic lesions were present in nearly 13% of these patients. At follow-up, 80% of these small lesions were determined to be benign and 8% were stable and indeterminate. Only approximately 12% of these liver lesions were confirmed as malignant. The high prevalence of benign liver lesions in adults, ^1,2 and the fact that small subcentimeter lesions are increasingly being recognized due to advances in CT and MR imaging, make liver lesion characterization an important objective of imaging. Incidental liver masses discovered in healthy adults as well as lesions detected during the work-up of a known malignancy frequently need to be characterized. Common benign lesions such as cysts, hemangiomas, focal nodular hyperplasia, focal fatty infiltration, and sparing need to be differentiated from metastases and HCC, and from less frequently encountered malignant lesions, such as fibrolamellar HCC, intrahepatic cholangiocarcinoma, lymphoma, mesenchymal tumors, and hepatoblastoma.

CT is often used as the initial modality for liver lesion characterization, and US may also detect and, occasionally, characterize lesions. Liver masses with US and/or CECT features typical of simple cysts or hemangiomas that are detected in patients who are not known or suspected of having a malignancy may confidently be classified as benign and generally do not need any further work-up. Similarly, lesions with unequivocal US and/or CECT signs of a malignant mass may not need additional imaging, but may sometimes need confirmation by percutaneous biopsy. For lesions that are indeterminate on US and CECT, MRI is the modality of choice to best characterize liver lesions (Figure 7).

LIVER LESION DETECTION AND CHARACTERIZATION ON MRI

When MRI first became a practical reality, it was thought that the administration of exogenous contrast agents would not be necessary and that native differences in relaxivity between lesion and normal liver parenchyma would suffice for both lesion detection and differential diagnosis. This was based on early studies that demonstrated that normal parenchyma and lesions possess different T1 and T2 relaxation times. These relaxation times could be used to differentiate the lesion from the parenchyma and, considering the difference in signal intensity between one lesion and another, could be used for the differentiation of these lesions. Specifically, some authors demonstrated that multi-echo heavily T2-weighted sequences are helpful in differentiating cysts and hemangiomas from metastases due to the different T2 relaxation times. ⁵⁰ Cysts and hemangiomas have longer T2 relaxation times than metastases and thus maintain high signal intensity on longer echo times (TE).

Although the information obtained on unenhanced T1- and T2-weighted imaging is frequently important for characterization, the potential overlap of signal intensity between benign and malignant lesions and between different lesion types within each group generally renders unenhanced imaging alone insufficient for accurate differential diagnosis unless a homogeneous, very high signal intensity lesion characteristic of cyst or hemangioma is encountered. This is particularly true with the now routine use of gradient-echo and fast-spin-echo techniques, rather than conventional T2-weighted imaging, which may lessen the differences in intensity between benign and malignant liver lesions. The limitations of the unenhanced approaches to MR imaging have made the use of contrast agents mandatory for adequate lesion characterization.

Contrast Agents for Liver MRI

As shown in Table 1, a number of contrast agents with different biodistribution patterns and relaxation properties are available today for liver MRI. Until relatively recently, the purely extracellular MR contrast agents based on gadolinium were the only agents available for both liver lesion detection and characterization. These T1 relaxing contrast agents are usually administered as a bolus and are used in conjunction with T1-weighted sequences such as T1-weighted gradient echo (GET1) or, more recently, volume interpolated breath-hold examination (VIBE). The lesion enhancement patterns observed derive from the differential perfusion properties of lesions compared with normal liver and are frequently specific to individual lesion types. Thus, lesion characterization is based essentially on differential lesion morphology and is therefore analogous to that occurring in helical CT.

In addition to the purely extracellular contrast agents, two distinct categories of MR contrast agents with liver-specific properties are now available (Table 1). The first category comprises the hepatocyte-targeted agents and include both gadolinium- and manganese-based agents (predominantly T1-shortening agents). The second category comprises the superparamagnetic iron oxides (SPIO), which are targeted to the reticuloendothelial system (predominantly T2-shortening agents). Agents of the former category include the gadolinium-based agents gadobenate dimeglumine (Gd-BOPTA; Multihance, Bracco Imaging SpA, Milan, Italy) and gadoxetate (Gd-EOB-DTPA; Eovist, Schering AG, Berlin, Germany), and the manganese-based agent mangafodipir trisodium (Mn-DPDP; Teslascan, Amersham-Health, Oslo, Norway). Agents of the latter category include Ferumoxides (Feridex IV, Berlex Laboratories, Wayne, NY; and Endorem, Guerbet, Aulnay Sous Bois, France) and SH U 555 A (Ferucarbotran, Resovist; Schering AG).

Unlike CT and MRI enhanced with purely extracellular gadolinium agents, MRI enhanced with agents with liver-specific properties is frequently able to make use of functional information to differentiate one lesion type from another. With Mn-DPDP and some of the larger SPIO agents (eg, ferumoxides) that cannot be administered as a rapid bolus, the functional information attainable is alternative to the morphologic information available with the extracellular gadolinium agents. With other agents, such as Gd-BOPTA, Gd-EOB-DTPA and some of the smaller ultrasmall superparamagnetic iron oxide agents (USPIO; eg, SH U 555A), rapid bolus administration is possible, and both morphologic and functional information is attainable.

Hepatocyte-targeted Agents

Mangafodipir trisodium is a manganese-based contrast agent that dissociates rapidly in the blood following slow intravenous infusion (2 to 3 mL/min over a 10- to 20-minute period) to release free Mn ⁺⁺ ion. This free Mn ⁺⁺ ion is then available for nonspecific uptake into parenchymal cells, particularly those of the liver, pancreas, kidneys, and adrenals in which metabolism of this metal takes place. Maximal enhancement of the liver is usually observed after approximately 20 minutes and lasts for approximately 4 hours. ⁵¹ Tumors of nonhepatocytic origin show little or no enhancement resulting in increased lesion conspicuity; several studies have shown improved lesion detection on images obtained after infusion of Mn-DPDP compared with precontrast images. ^52,53 Inherent in its uptake by liver cells is that it is also taken up by many hepatocellular lesions with well-differentiated hepatocytes resulting in enhancement, and thus a decreased tumor-liver contrast-to-noise ratio (CNR). An investigation aimed at evaluating Mn-DPDP for the study of hepatocellular tumors demonstrated poor efficacy of this agent for the identification of these lesions, in large part due to this phenomena. ⁵⁴ However, the uptake itself allowed for identification of additional lesions not seen on unenhanced imaging, but at the expense of converting other lesions clearly identified to isointense lesions. Moreover, Mn-DPDP accumulation has also been observed in hepatic metastases from nonfunctioning endocrine tumors of the pancreas. ⁵⁵ In a recent study in 77 patients with histologically confirmed lesions, Mn-DPDP­enhanced MRI had a sensitivity and specificity of 91% and 67%, respectively, for the differentiation of malignant versus benign liver lesions, and 91% and 85%, respectively, for the differentiation of hepatocellular versus nonhepatocellular lesions. ⁵⁶

Uptake of Mn-DPDP by both benign and malignant hepatocellular neoplasms limits the possibilities of accurate tumor differentiation and represents a major shortcoming of this agent. Moreover, there have also been reports of an increased toxicologic and neurologic risk in patients with hepatic impairment, and of a possible depressive action on heart function arising from the entry of free Mn ⁺⁺ into myocardial cells through calcium channels. ^57-59 Nonetheless, because of excellent liver parenchymal enhancement and virtually no uptake in common tumors, such as hepatic metastases, MR imaging with mangafodipir is now often used to replace CTAP for tumor detection and does allow for moderate tumor characterization.

Agents with Combined Perfusion and Hepatocyte-Selective Properties

Other contrast media combine the properties of a conventional extracellular agent with those of an agent targeted specifically to the liver. Such compounds distribute initially to the vascular-interstitial compartment in a manner analogous to that of the conventional gadolinium chelate agents. Thereafter, a fraction of the injected dose is taken up into functioning hepatocytes, causing an increase of the signal intensity of the hepatic tissue. Agents of this type include gadobenate dimeglumine (Gd-BOPTA), which is already approved in Europe and elsewhere for clinical use, and Gd-EOB-DTPA, which is in an advanced phase of development.

Gadobenate dimeglumine is the first of the combined (dual) second-generation gadolinium agents to become available commercially. ⁶⁰ Features that distinguish Gd-BOPTA from the conventional extracellular gadolinium agents are a capacity for weak and transient interaction with serum albumin, which results in a two-fold greater T1 relaxation rate in vivo, ⁶¹ and an elimination profile that sees roughly 96% of the injected dose excreted renally via glomerular filtration and the remaining dose taken up by functioning hepatocytes and eliminated in the bile via the hepatobiliary pathway. ⁶² Whereas the former feature permits lower overall doses to be used compared with conventional agents to obtain equivalent diagnostic information on dynamic imaging, ⁶³ the latter feature leads to a marked and long-lasting enhancement of the signal intensity of normal liver parenchyma, resulting in an additional, delayed imaging window beginning 40 min after Gd-BOPTA administration. ⁶⁴ Studies have shown that Gd-BOPTA behaves in an analogous way to conventional gadolinium agents during the dynamic phase of contrast enhancement, ^65,66 and as a liver-specific agent in the delayed phase when accentuated liver-lesion contrast differences not only improve the impact of MRI for the detection of focal liver lesions, ^66-68 but also contribute to the improved characterization of detected lesions, particularly lesions demonstrating atypical enhancement on dynamic imaging. ^69,70

Gadolinium-ethoxybenzyl-DTPA behaves in a similar manner to Gd-BOPTA in distributing initially to the vascular-interstitial compartment after injection. However, whereas only approximately 4% of the injected dose of Gd-BOPTA is thereafter taken up by hepatocytes and eliminated in the bile, approximately 50% of the injected dose of Gd-EOB-DTPA is taken up and eliminated via the hepatobiliary pathway after approximately 60 minutes. The maximum increase of liver parenchyma signal intensity is observed approximately 20 minutes after injection and lasts for approximately 2 hours. ⁷¹ During the perfusion phase, the dynamic enhancement patterns seen after injection of Gd-EOB-DTPA are similar to those seen with conventional agents and Gd-BOPTA. During the hepatobiliary phase, Gd-EOB-DTPA-enhanced images have been shown to yield a significant improvement in the detection rate of metastases, HCCs, and hemangiomas compared with unenhanced and Gd-DTPA­enhanced images. ⁷² It is not yet clear whether the additional hepatocytic uptake that this agent provides, as compared with Gd-BOPTA, will have any advantage in routine clinical practice.

Agents Targeted to the Reticululoendothelial System

Agents that distribute specifically to the reticululoendothelial system (RES) include the SPIO particles and the USPIO particles. These agents are cleared from the blood by phagocytosis accomplished by the RES, which is abundant in the liver parenchyma but is usually deficient or lacking in most malignant liver lesions. After SPIO injection, a darkening of the normal liver parenchyma surrounding focal liver lesions increases the CNR of these lesions, which, through having deficient RES, appear more hyperintense on delayed T2-weighted images compared with the decreased intensity of normal liver parenchyma.

Ferumoxides is an example of a larger type of SPIO agent with a diameter between 50 and 180 nm for which bolus injection and dynamic imaging are generally precluded. Thus, while this agent has proven effective for liver lesion detection, ^73,74 it has proven less effective for lesion characterization due to the overlap of diverse liver lesion types able to take up this agent. ⁷⁵

SH U 555A, on the other hand, is a smaller USPIO agent with a diameter between 45 and 60 nm. Unlike ferumoxides, this agent can be administered as a fast bolus to observe the early perfusion characteristics of the liver using T1- or T2*-weighted sequences. ^76-78 Unfortunately, the enhancement observed on SH U 555A­enhanced images is quite weak due to the small dose that can be injected.

AMI-227 is another example of a USPIO agent with a mean particle diameter between 17 and 20 nm. Because of the comparatively long intravascular half-life of this agent, AMI-227 is considered a blood-pool agent. ⁷⁹ As is typical of iron oxide contrast agents, AMI-227 shows preferential or selective accumulation in the liver parenchyma resulting in signal loss on T2-weighted images. With the long intravascular half-life, however, vascular lesions will also enhance for extended periods of time. This can aid in characterizing lesions, but may reduce liver-lesion contrast when compared with unenhanced images and may obscure some vascular tumors.

BEHAVIOR OF FOCAL LIVER LESIONS IN RESPONSE TO CONTRAST MEDIA

In the past, a practical approach to the characterization of focal hepatic lesions on CT and MR has focused on the distribution of conventional iodine and gadolinium chelates to the extracellular fluid spaces. This approach facilitates both the detection and characterization of lesions, the latter relying upon the recognition of distinctive patterns of enhancement of liver lesions. Lesions are broadly separated into categories of hypervascular or hypovascular on the basis of their handling of these contrast agents.

Hypovascular liver lesions include clearly avascular lesions, such as cysts and abscesses, and tumors that demonstrate only marginal peripheral enhancement, if any, compared with surrounding liver parenchyma. The majority of metastases to the liver are hypovascular lesions from primary sources such as colon, lung, pancreas, and others. The few primary sites that reproducibly demonstrate hypervascular liver metastases include renal, thyroid, neuroendocrine (pancreatic islet cell), and sarcomas. Breast and melanoma metastases to the liver may be vascular, but are generally less reproducibly so than the other tumors. Imaging of hypovascular tumors on CT or dynamic gadolinium-enhanced MR relies on imaging the liver during the peak liver enhancement, at which point liver-lesion contrast is maximal. Beyond this point, the CNR decreases as the equilibrium phases are reached (beginning at approximately 90 to 100 seconds postinjection). The hypovascular solid tumors, whether benign or malignant, all have a similar, noncharacteristic appearance and cannot be characterized further.

Hypervascular liver lesions include a wide number of lesion types ranging from physiopathologic to pathologic. These lesion types comprise benign and malignant lesions of both hepatocellular and nonhepatocellular origin. The enhancement characteristics of these lesions can aid in the differential diagnosis or, in some cases, can achieve a definitive diagnosis.

Benign Focal Masses Derived From Hepatocellular Origins

Focal nodular hyperplasia­ Focal nodular hyperplasia (FNH) is the second most common benign liver tumor, surpassed only by hemangioma. It is present in 3% to 5% of the population and is more common in women of child-bearing age. ⁸⁰ The cellular structure of FNH is similar to that of normal hepatic parenchyma apart from the presence of an abnormal biliary system. In more than a third of cases, the lesion develops asymptomatically as a result of a hyperplastic process ⁸¹ and is discovered in an entirely incidental manner. In roughly 85% of cases, only a single nodule of FNH is present. The margins of the lesion are usually well-defined, and its size is generally <5 cm, although lesions between 8 and 10 cm can be encountered. ⁸² Given the lack of clinical symptomatology associated with these lesions and the lack of malignant potential, FNH is most often left untreated, albeit with follow-up examinations to confirm the diagnosis. ⁸³

Since FNH contains the same elements as normal liver but with a disordered architecture, it mimics the appearance of normal liver and may be difficult to detect without exogenous contrast. On unenhanced MR imaging, the lesion is isointense to hypointense to normal liver on T1-weighted images, frequently with a hypointense central fibrous scar. On T2-weighted images, it is isointense to mildly hyperintense to normal liver with a hyperintense central scar. On dynamic gadolinium-enhanced T1-weighted images, the lesions show intense homogeneous enhancement in the arterial phase that washes out rapidly thereafter. This characteristic aids in differentiating these lesions from fibrolamellar hepatocellular carcinoma.

Typical of fibrous tissues, the central scar is slower to allow entry of extracellular-type contrast agents and slower to allow their washout. Thus, the scar appears characteristically hypointense on dynamic phase images, but iso- or hyperintense to the liver parenchyma on equilibrium and later images. This is frequently a distinguishing feature from the tumor necrosis often seen in malignant tumors.

Focal nodular hyperplasia lesions contain Kupffer cells and thus take up both SPIO and USPIO agents, demonstrating diminished signal intensity on post­iron-oxide scans. However, this finding is not specific, as well-differentiated HCCs may also take up iron oxides, although usually to a lesser degree. ⁸⁴ In equivocal cases with SPIO agents, biopsy or surgical resection may be required.

Although gadolinium-enhanced MRI is considered the most sensitive method for the characterization of FNH, atypical features are frequently seen. For example, in a recent study, 86% of small (¾3 cm) FNH did not have a visible scar on unenhanced or enhanced dynamic phase scans. ⁷⁰ Although the absence of a scar in small FNH cannot be considered wholly atypical, it may make it more difficult to distinguish these lesions from other hypervascular neoplasms on dynamic imaging alone. Hence, the availability of MR contrast agents with liver-specific properties, such as iron oxide agents, Gd-BOPTA, or Gd-EOB-DTPA, may be helpful for the accurate characterization of FNH.

Unlike the iron oxide agents, Gd-BOPTA and Gd-EOB-DTPA offer the possibility to perform both dynamic and delayed-phase imaging. The enhancement of FNH on dynamic imaging after Gd-BOPTA is indistinguishable from that after administration of conventional nonspecific gadolinium agents. However, on delayed T1-weighted images, substantial enhancement is noted within the parenchyma of the lesion, indicative of hepatocyte uptake (Figure 8), while the central scar, which is the principal site of biliary metaplasia, appears consistently hypointense. Similar findings have been observed with Mn-DPDP, ⁸⁵ although the inability to perform dynamic imaging with this agent is an obvious limitation.

Since Kupffer cells are usually observed within FNH, uptake of SPIO is common, with the lesions showing significant decreases in signal intensity on postcontrast T2-weighted images. ⁸⁶ Unfortunately, the amount and distribution of Kupffer cells within the nodules can vary, yielding different patterns of signal decrease: some small FNH (<3 cm) show homogeneous signal drop similar to that observed in the surrounding parenchyma, while large FNH may show heterogeneous signal drop. A recent intra-individual comparison of Gd-BOPTA and ferumoxides in 50 patients with 83 FNH revealed that while all 83 FNH (100%) were seen during the examination with Gd-BOPTA, only 62 FNH were seen during the examination with ferumoxides. Importantly, Gd-BOPTA detected, and was able to correctly diagnose, all 17 FNH in 12 patients with previous neoplasia, while ferumoxides was able to detect only 9 of these 17 lesions (52.9%) and accurately characterize only 7. ⁸⁷

Nodular regenerative hyperplasia­ Nodular regenerative hyperplasia (NRH) of the liver does not represent a specific entity but is a secondary and nonspecific tissue adaptation to heterogeneous distribution of blood flow, characterized by multiple monoacinar regenerative nodules without fibrous septa. These lesions can be imaged when the nodules achieve a macroaggregation of sufficient size to be visualized. Nodular regenerative hyperplasia occurs in 5% to 6% of individuals >80 years of age and with increased frequency in patients with Budd-Chiari syndrome, systemic arthritis, polymyalgia rheumatica, massive tumor infiltration, mineral oil deposition, and other disorders of liver blood flow. ^88,89

On unenhanced MRI, NRH displays a broad spectrum of signal intensity characteristics capable of appearing hypo-, iso-, or hyperintense with liver parenchyma on T1- and T2-weighted images. ⁸⁹ On dynamic-phase MRI with Gd-BOPTA and other extracellularly distributed gadolinium agents, approximately 90% of these lesions exhibit enhancement and appear hyperintense during the arterial phase and iso- or slightly hyperintense in the portal-venous and equilibrium phases. ^89-91 With both Gd-BOPTA and Mn-DPDP the lesions appear iso- or hyperintense to the surrounding liver parenchyma in the hepatobiliary phase. The appearance of NRH after SPIO administration has not yet been described completely, but due to the histologic characteristics of NRH, its behavior can be expected to be similar to that of FNH (Figure 9).

Hepatic adenoma­ Hepatic adenoma (HA) is a benign neoplasm typically found in women with a history of oral contraceptive use, ^92,93 and in patients with type I glycogen storage disease. ⁹⁴ In these latter subjects, the HAs tend to be multiple and more prone to malignant transformation, although this is still considered quite rare. ⁹⁵ Other predisposing conditions have been described, such as iron overload. ⁹⁶

Hepatic adenoma consist of plates or cords of cells that are larger than normal hepatocytes and contain large amounts of glycogen and lipid. The plates are separated by dilated sinusoids, which are thin-walled capillaries perfused by arterial pressure. A portal-venous supply is lacking in HAs. Kupffer cells are often found in HA but in reduced numbers and with little or no function, as reflected by the absent or diminished uptake of technetium (Tc)­99m sulfur colloid. ⁹⁷ Even though HAs have functioning hepatocytes, they lack bile ducts, a key histologic feature that helps distinguish HA from FNH. ⁹⁸ Thus, bilirubin metabolism is blocked within HAs, as confirmed by the absence of bile within resected HAs. ⁹⁹ Most patients are asymptomatic with normal liver function. The classic clinical manifestation for large or multiple adenomas is spontaneous rupture or hemorrhage leading to acute abdominal pain, and possibly hypotension or death. ⁹⁹ Thus, there is a clinical indication for the surgical removal of large (>5 cm) HAs.

The appearance of HA on unenhanced MR imaging has been variously described as hyperintense, isointense, and hypointense. ^100-102 Areas of increased signal intensity on T1-weighted MR images can result from fat and hemorrhage, ^103-105 while low signal intensity areas correspond to necrosis, old hemorrhage, or calcifications. ¹⁰³ Many HAs are predominantly hyperintense on T2-weighted images, while the remainder are isointense or hypointense. Most lesions are heterogeneous, demonstrating a combination of hyper- and hypointensity on T2-weighted images relative to hemorrhage and necrosis. One-third of HAs have a peripheral rim corresponding to a fibrous capsule, ¹⁰⁴ which is typically hypointense on T1- and T2-weighted images.

Dynamic gadolinium-enhanced MR imaging often reveals early arterial enhancement. On portal-venous phase images, this early enhancement usually fades, revealing an iso- or hypointense lesion. In lesions with previous hemorrhage, the arterial enhancement will be heterogeneous.

On delayed-phase images after injection of Gd-BOPTA or Gd-EOB-DTPA, there is little evidence of CM uptake by HAs which appear hypointense. ⁷⁰ On the other hand, the liver-specific contrast agent Mn-DPDP is able to enter the abnormal hepatocytes producing an iso- or hyperintense appearance on delayed-phase images (Figure 10). ⁵⁴ Due to the deficient Kuppfer cell activity, HAs usually do not show uptake of SPIO particles, resulting in an increased tumor-liver CNR ratio on T2-weighted images. However, on occasion, HAs may show some degree of uptake, ⁸⁶ leading to a heterogeneous signal drop similar to that seen in FNH. ¹⁰⁶ Uptake of SPIO in HA appears to be due to pooling of the contrast agent within the peliosis-like dilated vessels. ¹⁰⁷

Liver adenomatosis­ Unlike isolated HA, liver adenomatosis (LA) is distinguished by the presence of multiple adenomas, lack of correlation with steroid medication, occurrence in both men and women, and abnormal increases in serum alkaline phosphatase and gamma-glutamyltransferase levels, in an otherwise normal liver and in patients without glycogen storage disease. ^108-111

Histologically, the lesions consist of sheets of hepatocytes with foamy cytoplasm and uniform round nuclei. Portal tracts and bile ducts are absent, but Kupffer cells and fatty components can be present in some of the nodules. Many of the nodules can show evidence of subacute or chronic hemorrhage. Two different types of LA can be distinguished: a massive form and a multifocal form. ¹¹² Of these, the former has an increased risk of malignant degeneration and hemorrhage, and hence surgical intervention is frequently warranted. ^108,109

On T2-weighted MR images, most of the adenomas in LA are heterogeneously moderately hyperintense or homogeneously isointense. Noncomplicated lesions are generally isointense to the liver parenchyma on T1-weighted images.

Dynamic contrast-enhanced MR imaging with gadolinium contrast agents reveals most adenomas as hyperintense on arterial and early portal-venous phase images, and as isointense or homogeneously hypointense on late portal-venous and equilibrium phase images. On delayed-phase images after injection of Gd-BOPTA, most lesions appear hypointense, indicating a lack of uptake by the lesions. On the other hand, Mn-DPDP is able to enter the abnormal hepatocytes, resulting in a hyperintense appearance on delayed-phase images. Adenomas show variable uptake of SPIO, with lesion conspicuity decreasing at intermediate-weighted and T2-weighted imaging.

Hepatic cysts­ MR imaging is seldom required for characterization of hepatic cysts, but when performed, it reveals these lesions to be hypointense on T1-weighted images and hyperintense on T2-weighted images with the signal increasing markedly on heavily T2-weighted images. However, since hemangiomas also exhibit this behavior, a lack of enhancement of the cyst on postgadolinium T1-weighted images is able to distinguish between them if clinically indicated.

Malignant Focal Liver Lesions: Hepatocyte Origin

Hepatocellular carcinoma­ Hepatocellular carcinoma (HCC) is the most frequent primary tumor of the liver and represents >5% of all cancers worldwide, with an incidence of more than 500,000 new cases per year throughout the world. ¹¹³ There are considered to be three steps in the development of HCC: regenerative nodule, dysplastic nodule (low-grade; high-grade; with focus of HCC), and small HCC (<2 cm). ¹¹⁴ Regenerative nodules are benign lesions with exclusive portal-venous blood supply that occur nonspecifically in response to a variety of insults to the liver. Low-grade dysplastic nodules, on the other hand, show slight cytologic atypia, mainly large cell changes, while high-grade dysplastic nodules are considered pre-malignant lesions. The development of new nontriadal arterial flow to dysplastic nodules and small HCC can result in some enhancement on contrast-enhanced CT and MR and simulate HCC, although this is comparatively rare. ^115,116 In approximately one-third of high-grade dysplastic nodules, small foci of carcinoma can be found.

Dysplastic nodules are commonly encountered in severe cirrhosis, but only approximately 15% of such nodules are detected on MR imaging. ¹¹⁵ When visualized, dysplastic nodules are generally hyperintense on T1-weighted images and hypointense on T2-weighted images. ^117,118 Whereas arterial-phase enhancement can increase the sensitivity for characterization, this is at the expense of decreased specificity because of the overlap with the much more commonly seen arterial phase enhancement in HCC (Figures 11 and 12). ¹¹⁵ Although nodular regeneration and the development of dysplastic nodules increases the possibility of false-positive diagnoses of HCC, MRI is widely considered the most successful imaging modality for differentiating these nodules from HCC.

On the basis of histological differentiation, HCC can be classified as belonging to one of four grades, from I to IV. ¹¹⁹ Factors affecting the visualization of primary HCC during unenhanced and/or contrast-enhanced MRI of the liver include the dimensions, composition, and degree of vascularization of the lesion, the functionality of the normal hepatic parenchyma, and the residual hepatic functionality of the neoplastic cells themselves. Unfortunately, such factors vary from patient to patient, often making the behavior of a given HCC lesion difficult to predict. Studies aimed at correlating the appearance of HCC on MR imaging with the pathologic characteristics of the lesion reflect the difficulty in drawing firm conclusions on the behavior of such lesions. ^117,120-123

Many small HCC lesions are isointense on unenhanced T1- and T2-weighted imaging and, thus, are not easily seen. When visualized, HCCs are typically mildly to moderately hyperintense on T2-weighted images. On T1-weighted images, increased signal intensity correlates with a more well-differentiated histologic grade than isointensity or hypointensity. ^115,124 Dynamic T1-weighted imaging during the arterial phase improves the detection of small HCCs, which may be occult at other pulse sequences and on portal-venous and equilibrium phase images. ¹¹⁵

Frequently, larger HCC lesions display a pseudocapsule constructed from connective fibrous tissue around the tumor and internal septations. The pseudocapsule may derive from an interaction between tumor and host liver and may interfere with the growth and invasion of the HCC. ¹²⁵ The internal septation process (as well as heterogeneity inherent within larger lesions due to macroscopic fat accumulation, hemorrhage, necrosis, and fibrosis) creates a relatively characteristic mosaic appearance. ¹²⁶

The pseudocapsule of HCC appears hypointense on unenhanced T1- and T2-weighted MR images. On enhanced MR imaging with gadolinium-based agents enhancement is seen principally on portal-venous phase images, although late arterial-phase images may also demonstrate enhancement. Typical of fibrous tissue, enhancement persists into the equilibrium phase after washout from the normal liver parenchyma has begun (Figure 13). ¹²⁷

Kupffer cells are present in HCCs. Although there are no statistically significant differences in the numbers of Kupffer cells between small well-differentiated HCCs and noncancerous tissues, the numbers tend to decrease in cancerous tissue compared with noncancerous tissue as the tumor size increases and histologic grade decreases. ¹²⁸

While dynamic imaging of HCC with Gd-BOPTA provides similar information to that obtained with conventional gadolinium agents, delayed-phase imaging performed approximately 1 hour after contrast injection provides additional information for lesion characterization. Generally, moderately differentiated lesions tend to enhance to a greater extent on delayed images compared with both well-differentiated lesions and poorly differentiated lesions ⁶⁹ (Figure 14). This finding reflects the retention of sufficient hepatocytic activity by these lesions to take up Gd-BOPTA: A significant correlation was observed between the presence of intralesional bile and the degree of lesion enhancement after Gd-BOPTA administration. ⁶⁹

With the hepatobiliary agent Mn-DPDP, all hepatocellular tumors demonstrate enhancement, and HCCs frequently show heterogeneous en-hancement. ⁸⁵ Well-differentiated HCCs enhance to a greater extent than poorly differentiated HCCs. ¹²⁹

Post­Mn-DPDP MR imaging has been shown to result in a more accurate differentiation between benign (FNH) and malignant (HCC) hepatocellular tumors than unenhanced MR imaging, although absolute values for sensitivity, specificity, and accuracy were relatively low (64.5%, 66.7%, and 65.1%, respectively). ¹³⁰ A study intended to determine the efficacy of Mn-DPDP for the evaluation of HCC revealed minimal advantages compared with unenhanced MRI for the identification of lesions. ¹²⁹

With SPIO and USPIO agents, HCCs generally do not show a significant decrease in signal intensity, although signal intensity loss has been seen in some individual HCCs. ⁸⁶ The conspicuity of HCCs after SPIO depends on differences in the number of Kupffer cells between the lesion and the surrounding liver. Typically, moderately or poorly differentiated HCCs show large differences in the number of Kupffer cells and thus demonstrate a high CNR on SPIO-enhanced MRI, particularly in cirrhotic livers. Dysplastic nodules and most well-differentiated HCC, on the other hand, contain nearly the same number of Kupffer cells as the surrounding cirrhotic hepatic parenchyma and, therefore, are not well-depicted on T2-weighted MR images. ¹³¹ As with Mn-DPDP, the lack of a dynamic imaging capability limits the possibilities for lesion characterization with SPIO-enhanced MRI. With smaller USPIO agents, dynamic evaluation can be performed, but the conspicuity on enhanced images is generally weaker than with gadolinium-based agents (Figure 15).

Fibrolamellar HCC­ Fibrolamellar HCC (FLC) is a sharply defined, lobulated and nonencapsulated tumor, whose characteristic microscopic features include fibrolamellar bands of collagen and fibrocytes arranged in a lamellar pattern between nests of tumor cells, which often coalesce to form a central scar. ¹³² Typically, FLC occurs in a noncirrhotic liver, primarily in young adults, with no clear gender predominance. ¹³³

On MR imaging, the majority of tumors are hypervascular with a heterogeneous pattern. The heterogeneity of FLC permits its differentiation from FNH, another lesion with a fibrous central scar. The fibrous tissue within the scar and radial septa demonstrate persistent enhancement on contrast-enhanced MR images obtained 10 to 20 minutes after contrast administration. ¹³² On Gd-BOPTA-enhanced images, FLC tend not to show delayed enhancement at 1 to 3 hours postcontrast.

Benign Nonhepatocitic Focal Lesions

Hemangioma­ Hemangioma is the most common benign tumor of the liver, with a reported incidence of up to 20%. Although hemangiomas may be present at all ages, they are seen more commonly in premenopausal women. ¹³⁴ They are usually well-circumscribed, and represent tortuous caverns of blood-filled spaces, ranging in size from a few millimeters to >20 cm. Although frequently a solitary tumor, multiple hemangiomas occur in approximately 10% to 20% of cases. ¹³⁵ Frequently they occur in conjunction with FNH (in 15% to 20% of cases).

Marked hyperintensity on T2-weighted images is frequently diagnostic for hemangioma, although high signal intensity on heavily T2-weighted sequences with long echo delay times can sometimes be seen in cystic tumors and uncommonly in hypervascular metastases from sarcoma, islet cell tumor, pheochromocytoma, carcinoid, and renal cell carcinoma. On precontrast T1-weighted images, hemangiomas are commonly well-defined, slightly hypointense lesions with lobulated borders. Currently, gadolinium-enhanced MRI is the most sensitive technique for diagnosing hemangiomas. Three enhancement patterns of hemangioma with gadolinium-enhanced gradient-echo imaging have been noted: 1) immediate and complete enhancement of small lesions (so-called capillary hemangiomas); 2) peripheral nodular enhancement progressing centripetally to uniform enhancement; and 3) peripheral nodular enhancement with persistent central hypointensity due to central fibrosis. ¹³⁶ In the majority of cases the combination of T2-weighted and serial dynamic T1-weighted gadolinium images allows a confident diagnosis of hemangioma. ¹³⁷

Dynamic imaging characteristics of hemangiomas with Gd-BOPTA and Gd-EOB-DTPA are the same as with conventional gadolinium agents (Figure 16). ^62,65,66 On delayed-phase images after these agents and Mn-DPDP, hemangiomas tend to be hypointense compared with the surrounding liver parenchyma, ^56,66 as the delayed liver enhancement exceeds the eventual washout from the lesion blood pool.

Because hemangiomas do not contain Kupffer cells, no reticuloendothelial uptake is expected with SPIO imaging. However, hemangiomas do take up USPIO particles due to their blood-pool effect and may lose signal intensity on USPIO-enhanced T2-weighted images.

Malignant Nonhepatocytic Focal Lesions

Hypervascular Metastases-- Hypervascular metastases derive from highly vascular tumors such as carcinoid, islet cell tumor, renal carcinoma, thyroid carcinoma, pheochromocytoma, melanoma, and breast carcinoma. These lesions are usually hypointense on unenhanced T1-weighted images, and slightly hyperintense on T2-weighted images compared with background liver tissue. Frequently, these lesions demonstrate heterogenous signal intensity. Some hypervascular metastases may have higher signal intensity on T2-weighted images and thus mimic hemangiomas. ^138,139

Hypervascular metastases exhibit significant enhancement during the arterial phase after gadolinium injection, and as such can be distinguished from hypovascular lesions, which are best imaged during the portal-venous phase. In the portal-venous phase, hypervascular metastases usually show rapid peripheral contrast agent washout, which, on subsequent images, results in a hypointense halo appearance. ¹⁴⁰ Larger hypervascular metastases receive arterial blood mainly in the periphery of the lesion, and may thus demonstrate a peripheral rim of enhancement. ^138,139

The dynamic enhancement pattern of hypervascular metastases on Gd-BOPTA and Gd-EOB-DTPA­enhanced dynamic MR imaging is the same as that seen with conventional Gd-based contrast agents. However, in the delayed hepatobiliary phase, all metastases are seen as hypointense to the surrounding enhanced normal liver parenchyma (Figure 17). ⁶⁶ Although some contrast agent pooling may occasionally be observed in the necrotic center of the lesion on delayed images, repetition of the delayed sequence 2 to 3 hours after Gd-BOPTA injection, when complete washout from the interstitial space has occurred, is sufficient to confirm that this central enhancement is not due to hepatocytic accumulation.

As in the case of Gd-BOPTA, hypervascular metastases are typically hypointense to the surrounding liver parenchyma on delayed images after Mn-DPDP infusion. However, uptake of Mn ⁺⁺ after Mn-DPDP infusion has been observed in hepatic metastases from nonfunctioning endocrine tumors of the pancreas. ⁵⁵

On SPIO-enhanced T2- or T2*-weighted images, hypervascular metastases appear hyperintense compared with the low signal intensity of the surrounding normal liver parenchyma due to the lack of significant contrast agent uptake.

Hypovascular Metastases-- Hypovascular metastases are the most frequent malignancies in the liver. ¹⁴¹ In the United States, colon cancer is the most common primary site and approximately 50,000 cases of hepatic colorectal metastases are encountered annually. ¹⁴² Lesion detection is size-related; lesions <1 cm are difficult to identify with conventional techniques. ¹³ A postmortem assessment of the size of liver metastases has shown that the ratio between metastases >1 cm and those <1 cm is approximately 1:1.6 for metastases of colorectal adenocarcinoma and 1:4 for other liver metastases. ¹⁴³ Thus, an ability to accurately detect and characterize metastases <1 cm in size is clearly warranted.

Hypovascular metastatic lesions tend to have a round appearance, although large (>3 cm) metastases from colorectal adenocarcinoma commonly have a cauliflower aspect. The signal intensity of these metastases is lower than that of surrounding hepatic tissue on unenhanced T1-weighted images and typically moderately higher on T2-weighted images. In approximately 25% of colorectal hepatic metastases, a hypointense peripheral rim is observed in the parenchyma around tumor nodules on T2-weighted images. The associated histopathologic changes are compression of hepatic parenchyma, hepatocellular atrophy, fibrosis, inflammation, and congested sinusoids. ¹⁴⁴

It is currently a matter of debate whether conventional gadolinium contrast agents increase the detection of hypovascular metastases over unenhanced imaging. ²¹ On hepatic arterial-phase images, metastases can show a fleeting ring enhancement that blurs the margins of the lesions. This corresponds to a desmoplastic reaction, inflammatory infiltration, and vascular proliferation in the tumor-liver parenchyma margin; enhancement progresses centrally with concomitant peripheral washout. In the equilibrium phase, approximately 10 min after contrast administration, some lesions may show a peripheral hypointense rim (peripheral washout sign).

Dynamic-phase imaging after Gd-BOPTA administration reveals identical enhancement patterns to those obtained with conventional extracellular gadolinium-based contrast agents. However, since the lesions are unable to take up Gd-BOPTA, they appear as homogenously hypointense on delayed T1-weighted images (Figure 18). ^66,68 When there is a desmoplastic reaction to the lesion, an accumulation of contrast agent in the fibrotic part may be observed, which can remain for several hours; in these cases, the observation of a peripheral hypointense halo is highly suggestive of the malignant nature of the lesion.

A similar appearance is observed after Mn-DPDP infusion when the lack of contrast agent uptake leads to an increase of the liver-to-lesion CNR.

Since metastases do not contain Kuppfer cells, the CNR, and hence lesion detection and conspicuity, is improved after SPIO injection when compared with unenhanced T2-weighted images. ^145-147

Due to the nonspecific behavior of malignant liver lesions after SPIO injection, some authors have proposed a dual­contrast-agent study to improve lesion characterization. With this approach, pre- and post-SPIO images are obtained and followed, after bolus injection of a conventional Gd agent, by multiphasic dynamic T1-weighted gradient-recalled echo (GRE) images. ^148,149 The advantages of contrast agents such as Gd-BOPTA that have an inherent dynamic and delayed imaging capability may render the dual­ contrast-agent approach superfluous.

Cholangiocellular carcinoma-- Cholangiocellular carcinoma (CCC) is a malignant hepatic tumor of the biliary epithelium and is the second most common form of primary hepatic malignancy in adults after HCC. ¹⁵⁰ Nevertheless, it represents <1% of all newly diagnosed cancers in North America.

Cholangiocarcinoma is usually divided into intrahepatic and extrahepatic, depending on the site of origin. Intrahepatic cholangiocarcinoma (ICC) is a malignant neoplasm arising from the epithelium of the intrahepatic bile ducts and represents approximately 10% of all cholangiocarcinomas; hilar (Klatskin's) and bile duct cholangiocarcinoma account for the remaining 90%. ¹⁵¹ This neoplasm is usually a large, firm mass, and in 10% to 20% of cases there are several satellite nodules around the main lesion.

Intrahepatic cholangiocarcinoma has a nonspecific appearance on MR imaging. On precontrast T1-weighted images, it is generally iso- to hypointense relative to the normal liver, while on T2-weighted images, the signal intensity of the tumor ranges from markedly increased to mildly increased. Tumors with high fibrous content tend to have lower signal intensity on T2-weighted images, ^152,153 and those with a high mucin content very high signal intensity.

On serial dynamic T1-weighted images enhanced with conventional gadolinium agents, ICCs usually show minimal or moderate incomplete rim enhancement at the tumor periphery on arterial-phase images with progressive central contrast enhancement on portal-venous and early equilibrium phase images. ¹⁵⁴ Rarely, small homogeneously enhancing tumor nodules can be seen during arterial-phase imaging simulating HCC. Contrast enhancement on MR and CT is frequently better seen on delayed equilibrium images in patients with dense fibrous stroma. ^155,156

Imaging with Gd-BOPTA is similar to imaging with conventional nonspecific Gd-based contrast agents during the dynamic phase of contrast enhancement. However, delayed imaging with Gd-BOPTA during the hepatobiliary phase may reveal contrast enhancement in the fibrotic areas of the lesion. The degree of enhancement depends on the type of cholangiocarcinoma: poor enhancement is noted in large nonfibrotic cholangiocarcinomas (Figure 19), whereas greater delayed enhancement is noted in the hepatobiliary phase in the fibrous core of the scirrhous cholangiocarcinomas. ⁶⁶

No significant uptake of Mn-DPDP is observed on delayed T1-weighted images and, hence, ICC lesions generally appear hypointense although some peripheral rim enhancement may be observed. ¹⁵⁷ The hepatobiliary phase after administration of liver-specific contrast media may add useful information for identification of small satellite lesions.

Analogously, no significant uptake is observed after SPIO administration due to the absence of Kuppfer cells within the lesion. Consequently, there is increased liver-to-lesion CNR on T2-weighted images. ¹⁵⁸

CONCLUSION

Due to its physiologic location and function, the liver is a potential repository of a greater number and variety of focal lesions than any other organ in the body. Apart from malignant and benign tumors of primary hepatic origin, the liver is also a principal target for metastatic disease from malignant tumors elsewhere in the body. The role of diagnostic liver imaging, therefore, is not only to identify and quantify the presence of lesions, but also to differentiate lesions requiring immediate surgical or therapeutic intervention from lesions considered not to pose a significant clinical risk.

Among the diagnostic imaging modalities available, contrast-enhanced MRI is generally considered the most sensitive and accurate technique for the evaluation of focal liver disease. The variety of contrast agents now available permits accurate differentiation of benign from malignant lesions on the basis of morphologic information (ie, with conventional extracellular gadolinium agents), functional information (ie, with hepatobiliary and SPIO agents) or both (ie, with second-generation gadolinium agents, such as Gd-BOPTA).