Didier Mathieu is a Professor of Medicine and Radiology at the Centre d’Imagerie, Aix en Provence, France. The focus of his clinical research has been the diagnosis and treatment of liver tumours and the recognition and understanding of hepatic vascular disorders. He has published data on colour Doppler ultrasound, computerised tomography and magnetic resonance findings of benign liver lesions in Radiology, and their clinical management in Gastroenterology, the Lancet and the New England Journal of Medicine, from both multi-centre and national studies conducted in Europe and France.

Abstract
The development of ultrasound, computed tomography (CT) and magnetic resonance imaging (MRI) techniques, especially with the advent of specific contrast media, has improved the detection and diagnosis of liver metastases. Ultrasound (US) had a reputation for low sensitivity in the depiction of liver metastases, however, as a result of recent use of contrast media, it is now associated with improved detection rates and increased specificity. The introduction of multi-slice CT has allowed increased detection of multiple hepatic nodules and improved imaging of hepatic vascularisation. Magnetic resonance imaging is known for distinguishing liver metastases from other lesions, even with standard protocols and conventional contrast media. However, the development of hepatocyte-specific and reticulo-endothelial system-specific contrast media is producing even greater diagnostic accuracy. This article reviews the advantages and disadvantages of these techniques in the characterisation of hepatic metastatic lesions.

Introduction
The prevalence of liver metastases by far outnumbers that of primary malignant hepatic lesions. The prognosis of tumours such as colon carcinoma is directly related to the presence of distant metastasis, and this is especially true of the liver. Moreover, accurate lesion detection and localisation in the liver is mandatory for adequate treatment planning for a surgical resection or percutaneous treatment for residual liver tumours. Ultrasound, CT and magnetic resonance (MR) are standard techniques for liver imaging, and all have inherent advantages and disadvantages. The advent of contrast agents for US, the multi-slice technique in CT and hepato-specific contrast agents for MR has profoundly modified the role of each imaging modality. This article reviews these three imaging techniques and details the clinical impact of each on imaging in metastatic liver disease.

Ultrasound
Ultrasound is often considered as the first examination to be performed in patients with suspected focal liver lesions. However, the sensitivity of US for the depiction of liver metastases is low, ranging from 50-77%, mostly because of the inability of US to detect small lesions or iso-echoic metastases.² The presentation spectrum of liver metastases with US is thus wide and non-specific. Contrast agents have been recently introduced to improve lesion detection and characterisation.

Figure 1. Harmonic ultrasound before (a) and after (b) injection of sulphur hexafluoride (Sonovue®) (post-injection image was obtained in the late phase). Liver metastases are more clearly defined after a bolus injection than on plain examination

Micro-bubble contrast agents for US such as SHU 508A (Levovist™; Schering, Berlin, Germany) or sulphur hexachloride (Sonovue™; Bracco, Milan, Italy) improve vascular phase enhancement of Doppler signals. Furthermore, such agents are selectively taken-up by the normal liver in the delayed hepatic phase, allowing improved detection of focal liver lesions³. This delayed hepatic phase is observed 2-3 minutes following the contrast agent bolus injection, and may be related to a selective uptake by elements of the reticulo-endothelial system or to low-flow in the liver sinusoidal network.²

a

b

c

Figure 2. Computed tomography of hypervascular liver metastases from a renal primary tumour before contrast administration (a), and at the arterial phase (b) and equilibrium phase (c).

New US techniques have been specifically designed for use with micro-bubble contrast agents including pulse- or phase-inversion harmonic contrast-enhanced US.⁴ With such techniques, the US image is mainly created by the non-linear scatter produced by the contrast agent, while the background linear signal produced by the liver is cancelled, providing an optimal contrast-to-noise ratio of liver metastases over normal liver (Figure 1 on previous page). Large studies are still required to assess the exact role of contrast-enhanced US in the depiction of liver metastases, but the detection rate could reach that of helical CT.⁵

Computed tomography

Computed tomography remains the most frequently used technique for liver imaging following US.⁶ Although CT during arterial portography (CTAP) combined with helical CT hepatic arteriography (CTHA) provides high sensitivity,⁷ and remains for some the gold standard for depiction of liver metastasis,⁸ its invasiveness in routine practice has raised questions concerning its use. Furthermore, the advent of multi-slice CT (MDCT) has highlighted the clinical impact of helical CT for liver evaluation. Multi-slice CT enables the acquisition of multiple transverse images during a single gantry-rotation, thus allowing the scanning of large volumes within one breath-hold without decreasing the z-axis resolution.⁹ Although the exact increase in sensitivity and specificity obtained with MDCT over single- slice CT in liver metastasis assessment has not been precisely evaluated, MDCT has greatly improved the confidence of the radiologist in liver tumour detection (Figure 2). Moreover, MDCT allows imaging of hepatic vascularisation, by providing a multi-planar imaging capability.6 Such reconstruction techniques are believed to help treatment planning in patients with multiple hepatic nodules (Figure 3).¹⁰ For some authors, MDCT performed during a late arterial phase (with onset of acquisition 35 seconds after contrast media injection), could provide both liver lesion detection and proper vessel assessment.¹¹ Recently, Sahani et al. reported that the sensitivity and specificity of MDCT compared with conventional angiography for depiction of hepatic vessels were 94% and 100%, respectively.¹²

A standard protocol for liver imaging using MDCT comprises pre-contrast acquisition (with close to 2 mm collimation) followed by a triple-phase acquisition, including an early arterial phase (20 seconds after contrast injection), a late arterial or portal venous phase (35 seconds after contrast injection), and a hepatic venous phase (65 seconds after contrast injection). The optimal slice thickness is not known and varies depending on the CT unit selected. For some lesions, a slice thickness of 24 mm could be the most effective for detection.⁶ Despite these recent improvements however, the performance of MDCT and contrast-enhanced MR (especially with hepato-specific contrast agents) appear similar.¹³

Figure 3. Computed tomography 3-D reconstruction before surgical intervention

Magnetic resonance
As a result of the development of fast-acquisition techniques and improved gradients, MR is now routinely used for liver imaging, especially because of its ability to distinguish liver metastases from other benign lesions or from surrounding steatosis. The advent of liver-specific contrast agents, targeting either hepatocytes (such as manganese chelates) or Kupffer cells (such as ultra-small super-paramagnetic iron oxides [USPIO]) have further improved the diagnostic accuracy of MR. The subsequent sections examine the fundamentals of conventional liver MR using extracellular gadolinium chelates, and detail the impact of the more recently introduced liver-specific contrast agents.

Standard imaging protocols
Standard MR protocols for the liver usually comprise un-enhanced and contrast-enhanced sequences. Un-enhanced sequences combine T1-weighted images using breath-hold gradient echo (GRE) sequences with and without selective fat-suppression.¹⁴ They may be completed by in-phase and out-phase T1-weighted spoiled GRE sequences to improve lesion detection in subjects with a fatty liver, and followed by a breathing-averaged T2-weighted echo-train (such as turbo spin echo (TSE) or fast spin echo (FSE)) spin-echo MR sequences.¹⁵ Three enhanced acquisitions are performed following the dynamic injection of gadolinium chelates (hepatic arterial dominant, portal venous phase and hepatic venous or interstitial phase) and the signal-to-noise ratio is optimised by using 3-D acquisition sequences.¹⁶

Imaging features of liver metastases on MR with conventional extracellular gadolinium chelates
Liver metastases commonly appear hypo-intense on T1-weighted images and slightly hyper-intense on T2-weighted images.¹⁷ Areas of heterogeneity, especially with a markedly increased T2 signal intensity, are believed to correspond with areas of necrosis or cystic degeneration.18 A wide spectrum of enhancement patterns are reported with liver metastases. Hypo-vascular metastases include those from colon, bladder, prostate, and pulmonary carcinomas, whilst hyper-vascular metastases include those from tumours of the breast or thyroid, melanomas, carcinoid tumours, neuroendocrine tumours, and renal cell carcinomas.19 However, the most common pattern recently reported in a retrospective study of 516 lesions in 165 patients is that of peripheral rim-enhancement on arterial-phase acquisitions with incomplete progression of enhancement on delayed sequences regardless of the primary tumour site.17 Magnetic resonance is of significant impact in distinguishing metastases from benign lesions such as hemangiomas20 or from focal fatty infiltration.16 The efficiency of MRI in the detection of liver metastases has been recently assessed in a large meta-analysis: the overall reported sensitivity of conventional MR was 76% compared with 55% for US.²¹

Hepato-specific contrast agents
In order to improve the diagnostic accuracy of MR, hepato-specific contrast agents have been developed including reticulo-endothelial system (RES)-specific contrast agents (such as ferumoxides) and hepatocyte-selective contrast agents (such as manganese chelates, Gd-EOB DTPA and Gd-BOPTA).

Contrast agents targeting the reticulo-endothelial system
Iron oxides such as ferumoxides are selectively taken-up by elements of the reticulo-endothelial system (found in the liver, spleen, lymph nodes and bone marrow) which results in both a T2-shortening and a loss of signal intensity on T2-weighted images, especially using gradient echo techniques. Thus, the contrast-to-noise ratio of liver metastases (devoid of Kupffer cells) and normal liver (with Kupffer cells) is increased.¹⁵ Kumano et al. recently reported that the overall accuracy of SPIO-enhanced MR for differentiating benign-form malignant liver lesions was 96% using heavily weighted T1- and T2-gradient echo sequences. SPIO-enhanced MR also appears superior to both un-enhanced MR and spiral CT in the detection of liver metastases.²² Furthermore, the combination of SPIO injection, allowing a decrease in normal liver signal intensity, followed by gadolinium chelates injection, allowing an increase in contrast enhancement conspicuity, may not only improve the overall accuracy of MR on the detection of primary tumour,²³ but also may enhance detection of liver metastases.

Hepatocyte-specific contrast agents Hepatocyte-specific contrast agents are selectively taken-up by hepatocytes and account for a T1-shortening on structures showing contrast uptake.15 These contrast agents therefore allow an improved positive contrast-to-noise ratio between normal liver (increased signal intensity owing to hepatocyte capture of contrast) and liver metastases (devoid of hepatocytes and thus lacking contrast uptake and T1-shortening). Manganese chelates (mangafodipir trisodium, Mn-DPDP, Teslascan™) is to-date the most widely used hepatocyte-selective contrast agent over gadobenate dimeglumine (Figure 4).¹⁵ The diagnostic sensitivity and specificity of mangafodipir-enhanced liver MR may be equivalent to that of spiral CT.13 Just as with SPIO, a combined use of Mn-DPDP and gadolinium chelates could yield even higher diagnostic accuracy in the depiction of liver lesions.²⁴

Conclusion
As a result of the recent development of contrast agents for US, of multi-slice CT and of hepato-specific contrast agents for MR, liver imaging strategies in metastatic disease have been profoundly altered. Contrast-enhanced US, MDCT, and MR performed with hepato-specific contrast agents possess similar sensitivity and specificity in liver metastasis depiction. However, all these three techniques should be compared with an emerging imaging technique in the field of nuclear medicine: Positon Emission Tomography with 18 fluoro Desoxy Glucose (PET-DG).²⁵

References
1. Ohlsson B, Tranberg KG, Lundstedt C, et al. Detection of hepatic metastases in colorectal cancer: a prospective study of laboratory and imaging methods. Eur J Surg 1993;159:275–81
2. Albrecht T, Blomley MJ, Burns PN, et al. Improved detection of hepatic metastases with pulse-inversion US during the liver-specific phase of SHU 508A: multicenter study. Radiology 1993;227:361–70
3. Blomley MJ, Albrecht T, Cosgrove DO, et al. Improved imaging of liver metastases with stimulated acoustic emission in the late phase of enhancement with the US contrast agent SH U 508A: early experience. Radiology 1999;210:409–16
4. Burns PN, Hope Simpson D, Averkiou MA. Nonlinear imaging. Ultrasound Med Biol 2000;26, Suppl 1:S19–22
5. Quaia E, Bertolotto M, Forgacs B, et al. Detection of liver metastases by pulse inversion harmonic imaging during Levovist late phase: comparison with conventional ultrasound and helical CT in 160 patients. Eur Radiol 2003;13:475–83
6. Kopp AF, Heuschmid M, Claussen CD. Multidetector helical CT of the liver for tumour detection and characterization. Eur Radiol 2002;12:745–52
7. Kanematsu M, Hoshi H, Imaeda T, et al. Detection and characterization of hepatic tumours: value of combined helical CT hepatic arteriography and CT during arterial portography. AJR Am J Roentgenol 1997;168:1193–98
8. Matsuo M, Kanematsu M, Inaba Y, et al. Pre-operative detection of malignant hepatic tumours: value of combined helical CT during arterial portography and biphasic CT during hepatic arteriography. Clin Radiol 2001;56:138–45
9. Hu H. Multi-slice helical CT: scan and reconstruction. Med Phys 1999;26:5–18
10. Kapoor V, Brancatelli G, Federle MP, et al. Multidetector CT arteriography with volumetric three-dimensional rendering to evaluate patients with metastatic colorectal disease for placement of a floxuridine infusion pump. AJR Am J Roentgenol 2003;181:455–63
11. Francis IR, Cohan RH, McNulty NJ, et al. Multidetector CT of the liver and hepatic neoplasms: effect of multiphasic imaging on tumour conspicuity and vascular enhancement. AJR Am J Roentgenol 2003;180:1217–24
12. Sahani D, Saini S, Pena C, et al. Using multidetector CT for preoperative vascular evaluation of liver neoplasms: technique and results. AJR Am J Roentgenol 2002;179:53–9
13. Braga HJ, Choti MA, Lee VS, et al. Liver lesions: manganese-enhanced MR and dual-phase helical CT for preoperative detection and characterization comparison with receiver operating characteristic analysis. Radiology 2003;223:525–31
14. Semelka RC,Willms AB, Brown MA, et al. Comparison of breath-hold T1-weighted MR sequences for imaging of the liver. J Magn Reson Imaging 1994;4:759–65
15. Semelka RC, Helmberger TK. Contrast agents for MR imaging of the liver. Radiology 2001;218:27–8
16. Pedro MS, Semelka RC, Braga L. MR imaging of hepatic metastases. Magn Reson Imaging Clin N Am 2002 10:15–9
17. Danet IM, Semelka RC, Leonardou P, et al. Spectrum of MRI appearances of untreated metastases of the liver. AJR Am J Roentgenol 2003;181:809–17
18. Outwater E, Tomaszewski JE, Daly JM, et al. Hepatic colorectal metastases: correlation of MR imaging and pathologic appearance. Radiology 1991;180:327–32
19. Nino-Murcia M, Olcott EW, Jeffrey RB, Jr., et al. Focal liver lesions: pattern-based classification scheme for enhancement at arterial phase CT. Radiology 2000;215:746–51
20. Noone TC, Semelka RC, Balci NC, et al. Common occurrence of benign liver lesions in patients with newly diagnosed breast cancer investigated by MRI for suspected liver metastases. J Magn Reson Imaging 1999;10:165–69
21. Kinkel K, Lu Y, Both M, et al. Detection of hepatic metastases from cancers of the gastrointestinal tract by using noninvasive imaging methods (US, CT, MR imaging, PET): a meta-analysis. Radiology 2002;224:748–756
22. Reimer P, Jahnke N, Fiebich M, et al. Hepatic lesion detection and characterization: value of nonenhanced MR imaging, superparamagnetic iron oxide-enhanced MR imaging, and spiral CT-ROC analysis. Radiology 2000;217:152–58
23. Bhartia B,Ward J, Guthrie JA, et al. Hepatocellular carcinoma in cirrhotic livers: double-contrast thin-section MR imaging with pathologic correlation of explanted tissue. AJR Am J Roentgenol 2003;180:577–84
24. Martin DR, Semelka RC, Chung JJ, et al. Sequential use of gadolinium chelate and mangafodipir trisodium for the assessment of focal liver lesions: initial observations. Magn Reson Imaging 2000;18:955–63
25.Yang M, Martin DR, Karabulut N, et al. Comparison of MR and PET imaging for the evaluation of liver metastases. J Magn Reson Imaging 2003;17:343–49