This article discusses a study performed at Duke University on the advantages of using a high concentration CT contrast agent and multislice CT in liver imaging.
Rendon Nelson, MD
Professor and Vice-Chairman of Radiology, Duke University Medical
Center, Durham, NC
Faster multislice CT scanners create the opportunity to image
with increased speed, more reasonable breathhold times, better
image quality, and im-proved three-dimensional and multiplanar
Advances in CT technology also open the door to new approaches to
the administration of iodinated contrast. One possibility is the
use of higher-concentration, lower-volume contrast agents. Whether
such an approach could provide equivalent enhancement to that
achieved with standard-concentration contrast media is a question
we recently investigated at Duke University Medical Center.
We focused primarily on hepatic imaging. Therefore, a brief
review of both liver enhancement and key considerations in using
the new multislice CT technology may be helpful.
Balancing Speed and Image Quality
When upgrading from a scanner that acquires 4 slices per
rotation to one that acquires 8 slices per rotation, the first
option is to image much faster, for example, to reduce breathhold
times. With this approach, image quality remains about the same.
Another option is to image a little bit faster and enhance image
quality by decreasing noise and artifacts. The third option, when
acquisition and breathhold times are already acceptable with the
4-slice scanner, is to leave imaging speed unchanged and, instead,
attempt to achieve a data set with much higher image quality.
Radiation dose is also a consideration in selecting scanning
parameters. A 4 * 1.25-mm detector configuration uses a beam
collimation of 5 mm. On an 8-slice scanner, a slice width of 1.25
mm doubles the beam collimation to 10 mm, which raises concerns
about the radiation dose. When advancing from a 4-slice to an
8-slice scanner, it may be preferable to keep the beam collimation
approximately the same or use faster table speeds to prevent an
increase in radiation dose.
In liver imaging, lesion detection is determined by conspicuity,
which in turn is determined by liver-to-lesion contrast.
In tracking the phases of liver enhancement, the artery starts
enhancing approximately 15 seconds after contrast injection, and
then enhancement drops off a little bit. The portal vein starts
enhancing at about 30 seconds and peaks a few seconds after that.
The liver follows suit, and at about 60 seconds the liver
enhancement curve is characterized by an upslope and then a plateau
It is important to avoid imaging during the equilibrium phase,
which takes place at approximately the 3-minute mark. During this
time, lesions that can be clearly seen on the portal venous phase
tend to disappear (Figure 1).
Foley et al
characterized the phases of liver enhancement by scanning twice in
the arterial phase and once in the venous phase. They showed that,
for a hypervascular tumor, the enhancement curve tends to peak
during the later part of the arterial phase. At baseline, the
lesions may or may not appear as hypoattenuating masses. Many are
invisible, particularly if small. By about 20 seconds, contrast
fills the artery only, not having reached the hepatic parenchyma.
This early hepatic arterial phase is valuable for CT angiography,
but it is not very helpful for detection of hypervascular
At about 30 seconds, contrast is in the parenchymal arteries
and, to some degree, in the portal vein. This is what Foley calls
the portal venous inflow phase and others have referred to as the
late hepatic arterial phase. During this phase, a hypervascular
lesion will light up. At about 70 seconds, however, hypervascular
lesions can disappear. Foley called this the hepatic venous outflow
phase. It is also known as the portal venous phase (Figure 2).
In summary, in hepatic imaging, enhancement during the arterial
phase is predicated primarily upon the rate of injection, as well
as the timing. Timing can be determined by using a fixed delay, a
timing bolus, or automated triggering, which I prefer for
arterial-phase imaging of the liver and for CT angiography.
During the venous phase, however, enhancement is predicated not
so much on the timing or the rate of contrast delivery, but on the
total dose of iodine.
My colleagues and I at Duke University Medical Center decided to
further investigate this possibility with a randomized study.
The Duke Study: Methods
The hypothesis of the study was that a smaller dose of a more
highly concentrated iodinated contrast agent could be used to
obtain similar quality multislice helical CT images of the liver.
To test this hypothesis, we compared two contrast media injection
protocols, determining their liver and vascular enhancement
characteristics both quantitatively and qualitatively.
The prospective, randomized, double- blind study enrolled 20
patients with suspected or known liver lesions who were scheduled
to undergo abdominal CT. Our imaging parameters were as follows
(Table 1): For the non-contrast images, hepatic arterial phase, and
portal venous phase, we used a 5-mm collimation. We used a fixed
scan delay for both the arterial and venous phases, an approach we
debated, as this somewhat dated method does not reflect our
standard practice. Ultimately, we decided that we did not want
timing to be a confounding issue, since we were constructing time
attenuation curves. For the arterial phase, we selected a fixed
delay of 30 seconds, aiming for the later arterial phase. For the
venous phase, we selected a fixed delay of 65 seconds.
The table speed for the arterial phase was a little faster than
for the venous phase--22.5 mm/rotation, as compared with 15
mm/rotation--because the arterial phase is the shorter of the two.
Pitch also differed: It was 6-to-1 in the arterial phase and 3-to-1
in the venous phase. (It should be noted that to reduce the
radiation, we have recently changed our protocol for venous-phase
studies to specify a 4 * 2.5-mm detector collimation and a 6-to-1
pitch.) In the study, we also used a 0.8-second rotation time,
since not all of our scanners are capable of a 0.5-second gantry
The contrast injection protocol in one arm of the study
consisted of 150 mL of Isovue 300 (iopamidol, Bracco Diagnostics,
Princeton, NJ) delivered at 5 mL/sec, for a total iodine dose of 45
grams. That was compared with 100 mL of Isovue 370 delivered at 4
mL/sec for a total iodine dose of 37 grams. Therefore, the total
volume of contrast is reduced by one-third, while the iodine dose
is reduced by 18%.
When the iodine dose is analyzed by the number of grams
delivered per second, however, the difference is not substantial.
Administering 150 mL of Isovue 300 at 5 mL/sec delivers an iodine
dose of 1.5 gm/sec for 30 seconds. Administering 100 mL of Isovue
370 at 4 mL/sec delivers an iodine dose of 1.48 gm/sec for 25
We analyzed the images quantitatively, performing
regions-of-interest analyses on the liver in three places, and on
the aorta. We did this on every slice, and from that data we
constructed time attenuation curves. The images were analyzed
qualitatively as well by three abdominal radiologists who scored
them from 1 to 5--5 being the best--for liver enhancement, vascular
enhancement, and overall image quality.
In the liver, quantitative analysis of arterial-phase images
demonstrated enhancement ranging from approximately 13 to 15 HU
(Figure 3). En-hancement was a little greater with Isovue 370 than
with Isovue 300, but the difference was not statistically
In the portal venous phase, en-hancement was approximately 50 HU
after 150 mL of Isovue 300 and slightly less after 100 mL of Isovue
370, a not unexpected finding since, as noted above, total iodine
dose was 18% less with Isovue 370. However, the difference, again,
was not statistically significant (Figure 4). A
value of 0.12 raises the possibility of a statistical trend, but it
was not demonstrated in the 20 patients we studied.
In the aorta, there was more enhancement during the arterial
phase with Isovue 370, in the range of 250 to 270 HU (Figure 5),
but the differences between Isovue 300 and Isovue 370 were
This, however, was not the case during the portal venous phase.
For Isovue 300, enhancement in the aorta was approximately 100 HU,
while for Isovue 370, enhancement in the aorta was closer to 90 HU
(Figure 6), a statistically significant difference (
= 0.01). Again, the study design, which called for a shorter
injection, less volume, and less total iodine with Isovue 370,
makes this an expected result. Qualitative data, however, again
showed no difference between the two agents.
We concluded, therefore, that giving 100 mL of Isovue 370 at a
similar rate of iodine delivery and injection duration provided
comparable liver parenchyma enhancement in both the arterial and
the venous phases, when compared with 150 mL of Isovue 300.
There was a modest cost difference, however. Using the Isovue
370 protocol resulted in a 14% reduction in contrast use. That
could translate into an annual savings of nearly $54,000 at our
In adjusting CT scanning protocols for lower-volume,
higher-concentration contrast media, it is advisable to consider
the number of grams of iodine per milliliter, the rate of contrast
injection, and the duration of the injection and exam. In this way,
it is possible to tailor the protocol to achieve a similar rate of
iodine delivery per second as would be possible using standard
contrast media and, therefore, achieve a similar level of
enhancement in the target organ. *