1. Feridex: Ferumoxides (Feridex IV, Berlex Laboratories; and Endorem, Guerbet) are developed by AMAG Pharma (former Advanced Magnetics) and was referred to as AMI-25. The r2 and r1 relaxivites are 98.3 and 23.9 mM^-1sec^-1 respectively. Ferumoxides is available in USA, Europe, and Japan. Feridex is an SPIO colloid with low molecular weight dextran coating, with a particle size of 120-180 nm. To reduce the incidence of some side effects such as hypotension, Feridex is prepared as a dilution in 100 ml of 5% dextrose and administered as a drip infusion over about 30 min. At about 8 min following the intravenous injection, iron oxide particles are taken up by the reticululoendothelial cells in the liver and in the spleen with an approximate uptake of 80% and 6-10%, respectively (2). Maximum signal loss is obtained after 1 h with an imaging window ranging from 30 min to 6 h after the injection. The recommended dosage of Feridex IV (ferumoxides injectable solution) is 0.56 milligrams of iron (0.05 mL Feridex IV) per kilogram of body weight. Hypotension and lumbar pain/leg pain represent the most frequent symptoms associated with Feridex administration with an incidence ranging from 2 to 10%. Pain severe enough to cause interruption or discontinuation of the infusion was reported to occur in 2.5% patients.

2. Resovist : Ferucarbotran (Resovist, Bayer Healthcare) is developed by Schering AG, and was referred to as SH U 555A. Resovist is available in Europe and Japan. The active particles are carboxydextrane-coated SPIO, with a hydrodynamic diameter ranging between 45 and 60 nm. The r2 and r1 relaxivites are 151.0 and 25.4 mM^-1sec^-1 respectively. Unlike Feridex, Resovist can be safely injected rapidly in a bolus fashion, and the incidence of cardiovascular adverse events and back pain are significantly less. Resovist has an effect on the shortening of both T1 and T2 relaxation time. Resovist enables T1-weighted imaging ensuring a valuable although less pronounced positive T1 contrast effect. Dynamic T1-weighted GRE 3D sequences can be performed to acquire the perfusion properties of the lesion during the arterial and portal venous phases of the contrast agent. On dynamic MR imaging using T1-weighted GRE, enhancement was positive in the liver for at least 30 s after bolus injection of SPIO (3). However, positive enhancement of hypervascular hepatocellular carcinoma (HCC) in early phase of T1W-GRE has been reported to be weak to assess the tumor perfusion. Although this agent was found to cause significant T1 shortening of blood, its use for MR angiography was found to be suboptimal (4). Due to the high r2 relaxivity, Resovist is more suited to T2/T2*- weighted imaging. On delayed images after 10 min, the T2/T2* effects are observed due to the reticuloendothelial uptake in the liver. Perfusion study using echo planar imaging (EPI) yields negative enhancement of hypervascular tumors (5), and onestop shop diagnosis (involving both dynamic and RES-targeted MR imaging) for hypervascular HCC are feasible. Resovist come as 0.5 mmol Fe/ml solution in prefilled syringe. The recommended dose of Resovist is: for patients weighing less than 60 kg: 0.9 ml Resovist (equivalent to 0.45 mmol iron); for adults patients weighing 60 kg or more: 1.4 ml Resovist^® (equivalent to 0.7 mmol iron). Resovist’s overall incidence of adverse events was 7.1%, with vasodilatation and paraesthesia the most common event reported (<2%). Although considerably less post-marketing data is available on the safety of Resovist than on Feridex, the safety profile appears more favorable for Resovist.

3. Ferumoxtran-10 (AMI-227; Combidex, AMAG Pharma; Sinerem, Guerbet): The r2 and r1 relaxivites of Combidex/ Sinerem are 60 and 10 mM^-1sec^-1 respectively. The small size and hydrophilic coating result in a longer circulation in the intravascular space, and the particles escape rapid accumulation in the RES. These particles are phagocytosed by macrophages and accumulate in the lymphatic system. Normal lymph nodes are characterized by a dramatic signal drop on T2*- weighted images, whereas malignant lymph nodes, being devoid of macrophages, do not accumulate iron oxide particles and maintain a high MRI signal intensity. It takes 24 to 36 h for Combidex/Sinerem to accumulate in the lymph nodes, thus, postcontrast imaging is usually obtained 24 h after administration of the contrast agent. Sinerem was used in some European countries (not available now) — but not in the USA. In 2003, a paper by Harisinghani et al. (6) offered extraordinary results of the way in which ferumoxtran-10 could demonstrate the presence of positive lymph nodes in patients with prostate cancer. However, a recent, multi-center study by Heesakkers et al. (7) evaluated the use of ferumoxtran-10 and MRI to detect and identify lymph node metastases occurring outside the normal area of pelvic lymph node dissection in 296 patients with prostate cancer. All patients had intermediate to high risk for nodal metastases. There was a 24.1 % false positive rate in this study, leading to unnecessary surgical intervention. The clinical development for ferumoxtran-10 was stopped due to these results. Since ferumoxtran-10’ safety profile has been good as an imaging agent, perhaps the most appropriate question that researchers need to ask is where it is possible to develop a “second generation” form of ferumoxtran-10 which will have a significantly lower rate of false positive findings.

4. Clariscan: Clariscan (PEG-fero; Feruglose; NC100150) was developed by former Nycomed Imaging (now part of GE healthcare). NC100150 is consisted of SPIO particles that are composed of single crystals (4 to 7nm diameter) and stabilized with a carbohydrate polyethylene glycol (PEG) coat. The iron oxide particles have to be suspended in an isotonic glucose solution. The final diameter of an NC100150 particle is approximately 20 nm. Blood pool half-life is more than two hours in humans. It can be used as a MR angiography agent, and has been tested clinically for characterization of tumor microvasculature. NC100150 particles are eventually taken up by the mononuclear phagocyte system and distributed mainly to the liver and spleen. The development of NC100150 was discontinued due to safety concern.

SPIO-enhanced MRI is more accurate than nonenhanced MRI for the detection of focal hepatic lesions, and combined analysis of non-enhanced and SPIO-enhanced images is more accurate in the characterization of focal hepatic lesions than review of SPIO-enhanced images alone (9). In one early multicenter trial, Feridex -enhanced T2-weighted images revealed additional lesions not seen on unenhanced images in 27% of cases and additional lesions not seen by conventional (non-spiral) computed tomography (CT) scans in 40%; the additional information would have changed therapy in 59% of cases (10). The detection of metastases is apparently improved with SPIO agent, as well as cholangiocellular carcinoma, due to the absence of Kupffer cells within these lesions. Undifferentiated HCC usually demonstrate no change in signal intensity when compared with T2/T2*-weighted images in unenhanced and SPIO-enhanced imaging. This leads to an improvement in the contrast-to-noise ratio of the lesion with subsequent improvement of demarcation as well as visualization and an increased detection rate for HCC. On the other hand, lesions that contain reticuloendothelial cells, such as focal nodular hyperplasia, may become isointense to normal liver because of a decreased lesion-to-liver contrast ratio (11). A questionable focal nodular hyperplasia may be confirmed on SPIO enhanced MR. However, because of the relative inconsistency in the amount of reticuloendothelial cells in focal nodular hyperplasia and hepatic adenoma, cares should be taken with such clinical use.

Differentiation between HCC and dysplastic nodules (DN) is of great importance for the early and precise treatment of HCC in cirrhotic liver. One study reported that the ratio of the intensity of tumorous lesion to that of nontumorous area on SPIO-enhanced MR images (SPIO intensity ratio) correlated inversely with Kupffer-cell-count ratio in HCCs and dysplastic nodules, and increased as the degree of differentiation of HCCs decreased, indicating that the uptake of SPIO in HCCs decreased as the degree of differentiation of HCCs declined (12). Phagocytic activity might overlap among some borderline lesions. One study found no significant difference in number of Kupffer cells between well-differentiated HCC and surrounding liver tissue (13). Tanimoto et al. (8) reported that some welldifferentiated HCC exhibited signal decrease similar to the surrounding liver on T2W-FSE images, but less signal decrease than surrounding liver on T2*W-GRE images. Conversely, DNs exhibited strong decrease in signal on both T2W and T2*W images. In well-differentiated HCC, Kupffer cell density would be maintained but Kupffer cell function could be reduced compared to surrounding liver. One criterion, of a threshold signal loss of 10% on SPIO enhanced MR images, had been proposed to distinguish benign from malignant lesions (sensitivity 88%, specificity 89%) by receiver operating characteristic analysis (14).

The decrease in signal intensity of cirrhotic liver with SPIO is reduced compared to that in normal liver. The percentage of signal-intensity loss and liver-lesion contrast-to-noise ratio on SPIO-enhanced images was significantly higher in patients with mild liver cirrhosis than in patients with severe liver cirrhosis. Inflammation, scarring, regeneration and shunting in cirrhotic liver reduces hepatic uptake of SPIO, shifts distribution to the spleen, and produces signal heterogeneity (8,11). Structural and functional inhomogeneity in cirrhosis may cause false-positive lesions after SPIO administration.

To clarify the clinical role of SPIO-enhanced MR imaging in multi-modality decision-making, numerous comparative studies have been conducted. However, results drawn from such comparative studies should be carefully weighed since imaging equipment and parameters were not uniform among institutions. Investigators’ experiences and preferences might also play a role in the results. Final consensus has not been reached yet, or may not be reached due to consistent evolution of CT and MRI technologies.

1. SPIO-MRI versus dynamic CT: The combined approach of non-enhanced and SPIO enhanced T2-weighted MR images together resulted in a significantly higher sensitivity as well as in significantly more accurate differentiation of benign from malignant lesions as compared with results from spiral CT images, non-enhanced T2-weighted MR images or SPIOenhanced T2-weighted images alone (9). For the depiction of small hypervascular HCC, Lee et al. showed that the mean sensitivity of SPIO-enhanced MR imaging was significantly higher (70.6%, P<0.05) than that of dual-phase spiral CT (58.1%) (15). Tanimoto et al. (8) compared three imaging modalities in the detection of 72 HCCs. Detection rates were 69% for triple-phase dynamic CT (single helical), 89% for triplephase dynamic MR imaging, and 86% for SPIO-MR imaging. There was a significant difference among the three modalities in rate of detection of HCC (P<0.01), but not between dynamic MRI and SPIO-MRI. Kim et al. compared SPIO-enhanced MR imaging with triple-phase multi-detector CT (MDCT) for preoperative detection of HCC (). In their study, the mean sensitivities of MR imaging and triple-phase MDCT were 90.2% and 91.3%, respectively, and their mean specificities were 97.0% and 95.3%, respectively. SPIO-enhanced MR imaging was as accurate as triple-phase MDCT in preoperative detection of HCC (16). In addition, SPIO-enhanced MR imaging provides information supplementary to that obtained with dynamic CT, particularly by excluding pseudolesions. SPIO-enhanced MR imaging may be preferable due to its lack of radiation.

2. SPIO-MRI versus Gd-based dynamic MRI: Several studies have shown that Gd-based dynamic MRI is slightly better than SPIO-enhanced MR imaging in the detection of small HCCs (17,18). In lesion conspicuity, Gd-enhanced MR imaging is better than SPIO-enhanced MRI. However, SPIO yields additional information when imaging findings on Gdbased dynamic MRI are questionable because of intrahepatic arterioportal shunt (AP shunt) and/or post-therapeutic liver damage (19). Ward et al. (20) reported the usefulness of doublecontrast MR imaging, i.e. combined SPIO- and Gd-dynamic MR imaging, for diagnosis of HCC. SPIO-enhanced MR imaging (mean accuracy = 0.76) was more accurate than nonenhanced MR imaging (mean accuracy = 0.64, P<0.04), and double-contrast MR imaging (mean accuracy = 0.86) was more accurate than SPIO-enhanced imaging (P<0.05). Combined Gd-enhanced dynamic and SPIO-enhanced MR imaging may obviate the need for more invasive combined arterial portography and CT hepatic arteriography for preoperative evaluation of patients with HCC (8).