STAR ABSTRACT

Imaging the liver
selected topics

lawrence H
Schwartz

MD

Introduction
Many imaging modalities are used

to evaluate the liver. In the United
States, unlike many other parts of the
world, computed tomography is the
most commonly performed examina-
tion for evaluation of hepatic lesions.
MR! is used with increasing frequen-
cy in select clinical situations especial-
ly as MR! techniques are refined, and
MR! examinations are easier to per-
form. There are relative advantages
and disadvantages of both hepatic
MR and CT. In general, CT is less cost-
ly than MR, more readily available,
and most radiologists and many refer-
ring physicians have a relatively high
degree of confidence in looking at CT
images. Some studies, however, have
found that CT is less sensitive and spe-
cific than MR for detection and char-
acterization of focal hepatic disease.

Hepatic MR has several distinct
advantages over computed tomogra-
phy. MR provides outstanding intrin-
sic soft contrast that can enhance sub-
tle differences between normal and
pathologic tissues and tissues of dif-
ferent histologic subtypes. Non-ioniz-
ing radiation is used and MR! con-
trast agents are not nephrotoxic. MR!
images may be acquired with multi-
planar capabilities which are especial-
ly useful in depicting various anatom-
ic relationships. MR studies of the
liver take considerably longer than do

CT studies and are more difficult to
obtain uniform image quality.

There is still controversy about the
precise method of performing both
these studies. The issues of the type,
dose and delivery of IV contrast
media as well as equipment utilized
are under intense investigation. In
MR!, the pulse sequence used, pulse
sequence parameters and use of con-
trast is also under investigation.

The remainder of this talk will.
focus on selected issues in hepatic
imaging with CT and MR! as they
relate to oncology imaging.

Comparative
imaging studies

Initial studies
Many of the initial CT and MR!

comparative imaging studies per-
formed in the 1980s and 1990s used
dated technology that would not be
considered current now. Therefore,
the results of these studies are not
applicable with to day's imaging
equipment. Many of these studies
used axial CT instead of spiral or mul-
tidetector technology, therefore, thick
rather than thin slices were used,
which also limits comparison with
newer technologies.

MR technology was also relatively
slower in the past. This, again, result-
ed in studies that used thicker slices
and frequently used non-breath-hold
pulse sequences. In addition, none of
the currently used contrast agents
were available in these initial compar-
ative imaging trials, limiting both the

35 SA JOURNAL OF RADIOLOGY • December 2002

sensitrvity and specificity for the
detection of focal hepatic lesions.

Liver metastases
Metastases are the most common

liver malignancy and occur at least 20
times more frequently than primary
hepatocellular cancer. The evaluation
of metastatic disease to the liver is one
of the most common indications for
liver imaging. In general, liver MR is
more sensitive for the detection of
liver metastases than contrast-
enhanced CT.

Liver metastases have a wide vari-
ety of appearances on MR images.
Most are oflow signal intensity on T 1-
weighted images and bright on T2-
weighted images. Imaging features
that are suggestive of malignancy
include a target -sign or a halo of high
signal intensity peripherally, or a het-
erogeneous signal intensity, with ill-
defined borders. It is critical to differ-
entiate between benign hepatic lesions
including cysts and hemangiomas
and metastases. This distinction is rel-
atively easy with MR! and may be per-
formed with T2-weighted images and
contrast enhanced images.
Morphologically, on both T2 and post
contrast images metastases are gener-
ally complex and heterogeneous, and
frequently ill-defined, while cysts and
hemangiomas are homogeneous and
sharply defined.

Contrast-enhanced MRI, especial-
ly with gadolinium based agents,
increases both detection of liver
metastases and aids in liver lesion
characterization. Metastases have a
variable appearance after gadolinium
administration. They may be either
hypervascular and enhance on the
early arterial phase or hypovascular
and enhance later. They often enhance
with a complete peripheral ring. The



STAR ABSTRACT

use of MR contrast agents, other than
gadolinium DTPA, has been investi-
gated for a number of years and an
additional contrast agent has gained
FDA approval. Superparamagnetic
iron oxide (SPIO-ferumoxides) is a
reticuloendothelial specific, particu-
late MR contrast agent. SPIO's change
hepatic parenchymal contrast by
shortening the spin-spin relaxation,
resulting in a reduction in signal
intensity in tissue containing the con-
trast agent. Most hepatic tumors do
not contain reticuloendothelial cells.
Therefore, the contrast between the
tumor and normal hepatic parenchy-
ma will be increased. Ferumoxides
are principally a T2-contrast agent. If
it is necessary to do a Tl-weighted
sequence, then a preeontrast study
must also be performed. The current
formulation of this agent is adminis-
tered over 30 minutes by LV.infusion.
As this contrast agent and others enter
the market, a cost-benefit analysis will
need to be performed to assess the
added benefit, considering not only
the costs of contrast agent, but also
additional time required for imaging,
patient preparation and scan interpre-
tation.

Hepatocellular
•carcinoma

CT
Large hepatocellular carcinomas

tend to be heterogeneous, and may
demonstrate a typical mosaic appear-
ance on CT. Smaller HCCs are often
isodense and difficult to detect on
conventional CT, which is performed
during the portal venous phase of
enhancement. Conventional CT has a
sensitivity of 48% and a specificity of
70% in the detection of HCC. Non-
contrast and delayed images slightly

increase lesion detection by conven-
tional CT. The development of spiral
CT, which allows scanning during the
arterial phase of enhancement, has
been a major advance. Tumors can be
imaged during the period when many
HCCs are hyperdense relative to the
unenhanced parenchyma. Arterial
phase imaging detects 30 to 40%
more tumor nodules than conven-
tional CT, and will be the only phase
to show tumor in 7 to 10% of patients.

Direct CT arteriography
and CT arterioportography

Direct CT arteriography refers to
CT done during catheter injection of
contrast into the hepatic artery. HCCs
are detected as hypervascular lesions,
with a sensitivity of approximately
91%. This technique is invasive and
requires identification of accessory
arteries to the liver.It is also subject to
a variety of false positive results. CT
arterioportography (CTAP) refers to
CT scanning done during catheter
injection of contrast into the superior
mesenteric artery. Contrast flows to
the bowel and returns to the liver via
the portal vein, opacifying the normal
hepatic parenchyma. Liver tumors,
which are supplied by the hepatic
artery, appear as hypodense lesions
relative to the normal enhanced
parenchyma. CTAP is considered the
most sensitive preoperative method of
detecting liver tumors, but has several
limitations in cirrhosis. For example,
dysplastic nodules may be hypodense
and mistaken for HCC and perfusion-
al defects are common.

MR!
The morphology of HCC is well

demonstrated by MRI. Tumor cap-
sules and central scars are more fre-
quently seen than on CT. HCC

36 SA JOURNAL OF RADIOLOGY • December 2002

demonstrates variable signal intensity
on Tl-weighted images. Relative to
normal hepatic parenchyma, approxi-
mately 1/3 of HCCs are hypointense,
1/3 of HCCs are hyperintense, and 1/3
are isointense. High signal intensity
on Tl-weighted images is sometimes
due to the presence of intracellular
lipid, but in other cases the cause is
not known. Numerous studies have
investigated the characterization of
regenerative, dysplastic, and malig-
nant nodules by MRI. Contrast-
enhancement is important because
Tl and T2 signal intensity alone is
insufficient for reliable distinction of
these entities. Three different MRI
contrast agents have been studied;
gadolinium, ferumoxides, and man-
ganese. Gadolinium-DTPA is an
extracellular paramagnetic contrast
agent that produces enhancement in
vascular tissues on Tl weighted
images. Dynamic gadolinium-
enhanced MRI is the preferred
sequence for visualization of HCC,
because of the typical hypervascular
pattern of enhancement in the arteri-
al phase and because some well differ-
entiated HCCs may only be seen dur-
ing the delayed phase. HCCs do not
contain a significant number of retic-
uloendothelial cells, and so are more
easily visualized against the darkened
background on T2-weighted images.
While ferumoxides may help lesion
detection on T2-weighted images, fer-
umoxides do not appear to increase
lesion detection when compared with
gadolinium-enhanced MRI. Manga-
fodipir trisodium is a manganese
based hepatobiliary contrast agent
taken up by hepatocytes and secreted
in the bile. Manganese causes Tl
shortening. As a result, normal
parenchyma is bright on Tl W images,
and lesions are relatively dark.



STAR ABSTRACT

Three-dimensional
imaging

With the rapid scanning ability of
multidetector CT and volumetrie
MRI, it is feasible to obtain a three-
dimensional data set of the entire liver
during a single breath hold. With
reconstruction of this data, high-qual-
ity three-dimensional images may be
obtained. They are useful in presurgi-
cal planning and mapping of lesions.
These images may be combined with
other imaging of the liver to produce a
comprehensive, non-invasive method
for evaluating the hepatic parenchy-
ma, hepatic vasculature and biliary
tree.

Selected reading Comput Assist TOl11ogr1996; 20: 337-342.
7. Mitsuzaki K, Yamashita Y, Ogata I, et al.

1. Semelka RC, Worawattanakul S, Kelekis NL, et Multiple-phase helical CT of the liver for detect-
al. Liver lesion detection, characterization, and ing small hepatomas in patients with liver cir-
effect on patient management: Comparison of rhosis: contrast-injection protocol and optimal
single-phase spiral CT and current MR tech- timing. AfR 1996; 167: 753-757.
niques. J Magn Reson Imaging 1997; 7: 1040-

8. Oliver JH 3rd, Baron RL, Federle Mp, et al.1047.
Detecting hepatocellular carcinoma: value of

2. Hagspiel KD, Neidl KFW, Eichenberger AC, et unenhanced or arterial phase cr imaging or
al. Detection ofliver metastases: Comparison of both used in conjunction with conventional
superpararnagnetic iron oxide-enhanced and portal venous phase contrast-enhanced CT
unenhanced MRl at 1.5 T with dynamic CT, imaging. AIR 1996; 167: 71-77,
intraoperative US, and percutaneous US.

9. Earls JP, Theise ND, Weinreb JC, et al. DysplasticRadiology 1995; 196(2): 471-478.
nodules and hepatocellular carcinoma: thin-

3. Schultz JF,Bell JD, Goldstein RM, et al, Hepatic section MR imaging of explanted cirrhotic Uv-
tumor imaging using iron oxide MRl: ers with pathologic correlation. Radiology1996;
Comparison with computed tomography, din- 201: 207-214,
ical impact, and cost analysis. Ann Surg Oneol

10, Semelka RC, Lee JK, Worawattanakul S. et al.1999; 6: 691-698.
Sequential use of ferumoxide particles and

4, Semelka RC, Cance WGoMarcos HB, et aI, Liver gadolinium chelate for the evaluation of focal
metastases: Comparison of current MR tech- liver lesions on MRl, ] Magn Reson Imaging
niques and spiral CT during arterial portogra- J 998; 8: 670-674,
phy for detection in 20 surgically staged

11. Winston CB, Schwartz LH. Fong Y. et al.cases, Radiology 1999; 213: 86-91.
Hepatocellular carcinoma: MR imaging find-

S. Yamagami T. Arai y. Matsueda K, et al. The ings in cirrhotic livers and noncirrhotic livers,
cause of nonturnorous defects of portal perfu- Radiology. 1999; 210: 75-79.
sian in the hepatic hilum revealed by cr during

12. Lopez Hanninen E, Vagi Tl, Bechstein WO, et al.arterial portography. Am J Roentgenol 1999;
Biphasic spiral computed tomography for172: 397-402.
detection of hepatocellular carcinoma before

6. Stevens WR, Gulino SP, Batts KP, et al. Mosaic resection or orthotopic liver transplantation.
pattern of hepatocellular carcinoma: histologic Invest Radio11998; 33: 2J6-221.
basis for a characteristic CT appearance, J

~rimary CareRadiology--------I
FA Mettlet; Jt; Milton J Guiberteau/ Carolyn M Voss & C E Urbina
Here's a practical resource designed to help practitioners select appropriate radiologic tests for a full range of common disorders.
From simple x-rays to sophisticated imaging studies, this clinical reference examines virtually every procedure in use today. It's
organized by anatomic system, with a focus on symptoms and suspected clinical problems, to make reference simple. This valuable
tool helps readers order even the most expensive and complex radiologic studies with confidence.
Features
e:> Examines the pros and cons of a broad spectrum of imaging studies, including their diagnostic value and their cost effectiveness.
e:> Addresses when a study should or should not be ordered, and the proper sequence in which radiologic tests should be

performed for specific symptoms or suspected disorders.
e:> Offers full coverage of common inpatient and outpatient conditions, such as chest pain, asthma, diabetes, and back pain.
e:> Describes both adult and pediatric conditions.
e:> Illustrates normal anatomy, normal variants, and common diseases and disorders with more than 300 clearly reproduced illustra-

tions.
e:> Includes step-by-step guidance on the interpretation of x-ray images.
e:> Considers common pitfalls and how to avoid them.
e:> Highlights vital information with quick reference tables.
e:> Integrates the perspectives and experience of two radiologists, a general medical internist, and a family practice physician.
May 2000, paperback, 285 pp, 340 illustrations, WBS, R799

Orders: SA Medical Association, Private Bag Xl, Pinelands 7430. Tel (021) 530-6527.
Fax (021) 531-4126. E-mail: fpalm@samedical.org Limited local stock.

37 SA JOURNAL OF RADIOLOGY. December 2002

mailto:fpalm@samedical.org