CTA is changing the evaluation of both coronary artery disease and pulmonary emboli. Advanced scanner technology is making it possible to do both examinations in a single scan, vastly improving the evaluation of patients with chest pain.
is a Professor of Radiology and Director of Thoracic Radiology at
the University of Michigan Medical Center, Ann Arbor, MI.
Cardiovascular disease is the leading cause of death in the
United States, responsible for nearly 40% of deaths each year.
The combined direct and indirect costs of cardiovascular disease
exceed $393 billion annually. Of the lives lost to cardiovascular
disease, coronary heart disease claims 53%, making it the leading
cause of cardiovascular death. In fact, in 2002, nearly a
half-million Americans died of coronary heart disease.
The gold standard test for the diagnosis of coronary artery
disease (CAD) is cardiac catheterization. Though effective, this
procedure is invasive, expensive, and associated with morbidity and
mortality. According to 1990 registry data from the Society for
Cardiovascular Angiography and Interventions, the total rate of
complications during diagnostic cardiac catheterization was 1.77%,
including death, myocardial infarction, arrhythmia, neurological
complications, vascular complications, and adverse reactions to
In addition, the accepted rate for normal findings on cardiac
catheterization, defined as insignificant or absent luminal
narrowing, is 20% to 27%.
A noninvasive test that could identify patients with normal
coronary arteries would be expected to reduce the number of
As little as a decade ago, noninvasive alternatives barely
existed. Today, they are hailed in the popular press as
breakthroughs in the evaluation of patients with cardiovascular
disease. In January 2004, for example,
The Wall Street Journal
placed coronary computed tomography (CT) and magnetic resonance
imaging (MRI) at the top of its list of 10 major medical advances.
The article, "A Better View of Your Heart,"
noted that rapid advances in CT and MRI technology are changing the
rules of angiography, making the tests available not only to
patients with symptoms, but also to those with worrisome risk
Given rapid advances in technology, coronary CT angiography
(CTA) may soon become a routine part of clinical practice.
Previously, with single-, 4-, and 8-row scanners, coronary CTA
demonstrated variable degrees of success in detecting critical
stenoses, and, as a result, was not considered robust enough for
everyday clinical use.
That consensus began to shift with the introduction of 16-row
multidetector CT (MDCT) technology. The ability to acquire data in
submillimeter-thick slices (0.625-mm or 0.75-mm) enabled isotropic
imaging in the z-axis and improved spatial resolution. At the same
time, faster gantry rotation speeds (0.4 seconds) improved temporal
resolution, reducing motion artifacts associated with the beating
Several studies have examined the ability of 16-row coronary CTA
to detect CAD. Nieman et al
studied 59 patients who were scheduled to undergo elective cardiac
catheterization, excluding those with an irregular heart rate.
Using a 0.75-mm detector collimation and a 0.42-second gantry
rotation, investigators evaluated coronary arteries ≥2 mm in
diameter for stenoses of at least 50%. They found that coronary CTA
had a sensitivity of 95% for detecting critical stenoses and a
negative predictive value of 86% among the evaluable coronary
artery segments. The overall accuracy of coronary CTA depended on
the vessel. It was very high in the left main and the left anterior
descending coronary arteries (100% and 91%, respectively) but was
lower in the right and circumflex coronary arteries (86% and 81%,
A similar study by Ropers et al
enrolled 77 patients scheduled to undergo cardiac catheterization,
excluding those with cardiac arrhythmias and unstable angina. Using
a scan protocol similar to that of the Nieman study, the
investigators examined coronary artery branches ≥1.5 mm in diameter
for stenoses of at least 50%. Coronary CTA demonstrated a
sensitivity of 92% and a negative predictive value of 97% among the
evaluable coronary artery segments.
However, these studies do not yet provide enough information to
determine the role of coronary CTA in clinical practice. A
consistent finding among studies of 16-slice coronary CTA is the
large number of nonevaluable coronary artery segments. In the study
by Ropers et al,
38 of the segments involving 12% of patients were not evaluable
because of excessive coronary calcification or motion
Table 1 outlines the technique we use at the University of
Michigan for 16-row coronary CTA. Scanning the entire heart from
apex to base (or up through the subclavian arteries, in the case of
a bypass graft) takes 15 to 29 seconds, depending on the scanner
platform, patient size, detector collimation, and gantry rotation
Since delivery of our 64-row volume CT (VCT) scanner, we use it
to perform all coronary CTA. The technique is similar to that
described in the 16-row protocol, but data acquisition takes only 5
to 6 seconds, resulting in a more robust dataset and more
consistent evaluation of coronary arteries as small as 1 to 2 mm in
We use a detector collimation of 0.625 mm, an overlapping narrow
pitch (0.2 to 0.35, depending on heart rate), and the fastest
gantry rotation possible on our scanner (0.42 seconds). We
administer 80 mL of iodixanol (Visipaque, GE Healthcare, Princeton,
NJ) 320 mgI/mL, injecting at 4 mL/sec. The amount of contrast
material depends on the scan time, the patient's individual
physiology, and the scan range. We use a 30- to 50-mL saline chaser
bolus to keep the contrast column moving and reduce overall
contrast volume. With the VCT, we have routinely reduced our
contrast volume to 80 mL.
At the beginning of each examination, we do a
non-contrast-enhanced, electrocardiographically gated acquisition
to assess the degree of coronary calcification and to map the
coronary arteries. The noncontrast acquisition confirms that we
have selected the right scan range for the heart and the coronary
arteries. This step is particularly important in patients who have
undergone coronary artery bypass grafting, as we need to identify
the graft's origin before the scan to ensure that it is included in
the field-of-view during coronary CTA.
To optimize contrast enhancement, we deliver a 20-mL timing
bolus prior to image acquisition, documenting a time-density curve
and using the time to peak enhancement, plus 5 seconds, as the scan
delay. This delay provides enough time for contrast material to
opacify the distal coronary artery branches before image
Speed and motion control are critical to imaging moving
structures, particularly the coronary arteries. Cardiac motion
artifacts can be minimized in several ways. One is through faster
scanning, made possible by newer-generation CT scanners. In
addition, heart-rate control remains a key step in coronary CTA.
The faster the heart rate, the more numerous the image artifacts
and the less accurate the detection of stenoses. The heart rate can
be manipulated through not only the use of beta-blockers but also
the selection of contrast material.
An early study by Schroeder et al
illustrated the impact of heart rate on vessel visibility. Using a
4-detector-row CT scanner, these investigators found that the
patients with the highest number of visible coronary segments had
the lowest heart rate, a mean of approximately 60 bpm. They also
showed that segment visibility was inversely correlated with heart
rate (r = 0.48;
A study of 100 patients by Gerber et al
showed that heart rate influenced the number of coronary segments
with motion artifacts. Patients with a heart rate of 51 to 60 bpm
had motion artifacts in only 12% of coronary segments, whereas
patients with a heart rate of 61 to 70 bpm had motion artifacts in
71% of the segments.
Heart rate can impact not only image quality but also the
ability to detect stenoses. In a study of 100 patients who
underwent coronary CTA on an 8-row scanner, Giesler et al
showed that the sensitivity for the detection of critical stenoses
was 62% when the heart rate was <70 bpm, but only 33% at higher
Most physicians who perform coronary CTA use a beta blocker to
reduce the heart rate to <65 bpm. At the University of Michigan,
we use a combination of oral and intravenous metoprolol, depending
on the patient's heart rate just before the scan. Occasionally, in
patients with very rapid heart rates in whom we need very tight
control, we administer intravenous esmolol, a very-short-acting
We also use isosmolar contrast medium for all coronary CTA
examinations as a way of minimizing fluctuations in heart rate and
blood pressure. Studies by Roriz et al
and Tveit et al
have shown a greater increase in heart rate with the low-osmolar
agent ioxaglate than with the isosmolar agent iodixanol.
An increase in heart rate in response to the administration of
contrast medium is more common in patients with severe heart
disease, including coronary artery stenosis or prior myocardial
It may occur due to a reduction in blood pressure or to patient
stress and discomfort. These and other studies have also shown that
iodixanol is associated with fewer patient reports of discomfort
and warmth when compared with low-osmolar contrast material.
Rather than assigning technologists to do image postprocessing,
physicians at our institution interact with the data sets,
reconstructing images as part of the interpretation. We generate
views of the coronary arteries in multiple planes. Figure 1A shows
a curved reformat of the right coronary artery, visible down to a
diameter of 2 mm and smaller. Figure 1B shows the left main
coronary artery as it branches into the left circumflex and left
anterior descending coronary arteries.
We can also generate surface displays, add color, and create
stunning images to send to clinicians and patients. As shown in
Figure 2, the result is more than artistry. This surface display
depicts very fine anatomic detail in vessels as small as 1 to 1.5
mm in diameter.
Coronary CTA can evaluate the artery in ways that conventional
cardiac catheterization cannot, depicting not merely narrowing of
the arterial lumen, but also remodeling and external expansion of
the arterial wall. Figure 3 shows a 16-row CTA in a middle-aged
hypertensive patient with a family history of coronary artery
disease. His coronary calcium score was zero, but contrast-enhanced
coronary CTA showed a single, noncalcified, focal eccentric
thickening of the coronary artery wall-a soft plaque. Cardiac
catheterization confirmed the location of the lesion, and
intravascular ultrasound confirmed that it was a noncalcified
There is more to cardiac CT than evaluation of the coronary
arteries. These powerful volumetric data sets contain an enormous
amount of information that we can view in very high resolution in
any plane. Short-axis, long-axis, 3-chamber, and 4-chamber views
are standard in cardiology, whether in echocardiography, nuclear
medicine, cardiac MR, or cardiac CT. Figure 4 shows a short-axis
view of the right and left ventricles. In the long-axis view of the
left atrium and left ventricle, it is possible to see the mitral
valve with its leaflets open, as well as the papillary muscles. The
3-chamber view depicts with exquisite detail the mitral valve with
its leaflets open and the aortic valve closed during
By adding the element of time to the interpretation of the data
set, it is possible to view the heart as it moves through the
cardiac cycle. This enables the evaluation of regional and global
wall motion abnormalities, as well as quantification of cardiac
function through documentation of left ventricular mass,
end-systolic and end-diastolic volumes, stroke volume, and ejection
Role of coronary CT
It is not yet clear what role coronary CTA will play in the
evaluation of patients with suspected CAD. Its potential as a
screening test must be compared with existing alternatives, such as
conventional treadmill and nuclear stress testing. In the future,
however, coronary CTA might be used to screen individuals with
multiple risk factors and to identify critical lesions.
Coronary CTA may also reshape the diagnosis of CAD and redefine
the criteria for vascular intervention. Today, a stenosis of ≥70%
is considered to be physiologically significant, warranting bypass
grafting or percutaneous coronary intervention. The ability of
coronary CTA to detect remodeling and external expansion of the
arterial wall raises several new questions. Where does one measure
the true coronary artery diameter? How does one define a critical
lesion? For example, should a stenosis of 55% in association with
soft plaque prompt aggressive treatment?
CT also has the potential to evaluate cardiac function in
patients already undergoing coronary CTA. The increased coverage
made possible by 64-row scanners may also lend CT the ability to
evaluate myocardial perfusion and viability.
Perhaps the most exciting future application of coronary CTA is
in plaque characterization and quantification. Closely linked is
its potential to guide drug therapy. Patients treated with
aggressive lipid-lowering may undergo serial coronary CTA
examinations, first to quantify the baseline plaque burden and then
to determine whether lipid-lowering medications are effective in
reducing that burden.
Pulmonary embolism (PE) is difficult to diagnose. Physical signs
and symptoms are nonspecific, and conventional diagnostic tests are
insufficiently sensitive and specific. As many as 70% of
ventilation-perfusion (V/Q) scans are nondiagnostic, for example.
Even catheter pulmonary angiography, despite being considered the
"gold standard," is imperfect. A 1990 study showed that pulmonary
angiography had a sensitivity of only 87% for PE.
Catheter pulmonary angiography is also underutilized. Studies by
Schluger et al
and Sostman et al
found that, in the majority of cases, physicians chose not to
perform pulmonary angiography if the V/Q scan was
For these and other reasons, pulmonary CTA is replacing
conventional tests. Pulmonary CTA appears to be comparable to
invasive angiography for the detection of PE. In a porcine model,
Baile et al
reported that pulmonary CTA had a sensitivity of 82% and a
specificity of 94% for the detection of pulmonary emboli in 1-mm
subsegmental vessels. Invasive angiography, by comparison, had a
sensitivity and a specificity of 87% and 88%, respectively.
Pulmonary CTA has substantial advantages over V/Q scanning. It
enables direct imaging of the thrombus. When combined with indirect
CT venography, it identifies other diseases in 25% to 50% of
patients. It is highly predictive of patient outcomes.
Specifically, a negative pulmonary CTA is associated with a low
incidence of PE and deep-vein thrombosis during the subsequent 3 to
12 months. Finally, its interobserver agreement is higher than that
of alternative tests, such as the V/Q scan and catheter angiogram.
Each of these advantages will be discussed here in more detail.
Figure 5 illustrates CTA detection of emboli in the pulmonary
arterial circulation. Figure 5A depicts large central pulmonary
emboli, and Figure 5B shows nonenhancing right lower lobe
atelectasis, which can lead to finding clots that might otherwise
have been overlooked.
In critically ill ventilator-dependent patients, pulmonary CTA
is technically challenging and is often characterized by streaking
and image noise. Still, it is possible to identify pulmonary emboli
on these images, not just centrally, but even in smaller segmental
and subsegmental arteries (Figure 6).
One of the key advantages of CT is that it can detect venous
thrombi in the pelvis, legs, and upper extremities, thereby
identifying the source of pulmonary emboli and determining whether
there are remaining venous thromboemboli that could migrate to the
systemic circulation. Detecting clots in the common iliac,
infrapopliteal, and other upper- and lower-extremity veins through
use of a combined CT angiographic/venographic study can improve
diagnostic yield by about 20%, as compared with CTA alone.
CT can also detect other disease processes responsible for the
clinical presentation in as many as 50% of patients.
These could include progression of lung cancer, mediastinal
fibrosis, or infectious pericarditis, each of which could mimic the
signs and symptoms of PE. In addition, CT can reveal other,
unrelated findings that are important in the patient evaluation-for
example, right heart failure with tricuspid valve incompetence.
The interobserver agreement of pulmonary CTA is better than that
of V/Q scanning and probably equal to that of catheter angiography.
Pulmonary CTA is also highly predictive of patient outcome. An
analysis of 8 studies with a combined enrollment of nearly 2000
patients showed that only 1% of patients with a negative pulmonary
CTA examination actually developed deep-vein thrombosis or PE
during follow-ups of up to 1 year.
Thus, a negative CTA study appears to be as predictive of outcome
as a negative invasive pulmonary angiogram.
The technique used in performing pulmonary CTA has changed over
time, along with a rapid evolution in scanner technology.
Acquisition times have gone from 30 seconds to 3 seconds. Coverage
has improved, from 12 cm-encompassing only the central part of the
pulmonary arterial tree-to more than the 25 to 30 cm needed to
cover entire thorax. Collimation has dropped from 5 mm to 0.5 mm.
In addition, we can now easily add indirect CT venography to the
Advancing technology is prompting an important clinical change
as well. With today's 64-slice CT scanners, we are exploring the
use of a so-called triple rule-out study for patients with chest
pain. In such patients, CTA can determine whether the chest pain is
due to coronary artery disease, aortic disease, or pulmonary
thromboembolic disease. With 64-slice scanners, it is possible to
obtain all of this information in one scan, as long as the imaging
technique is tailored appropriately.
There is an ongoing debate over the best approach to contrast
administration in pulmonary CTA. One camp advocates using a fixed
scan delay for every patient; another camp advocates tailoring the
scan delay to each patient's physiology.
A fixed scan delay is easy and works well in the majority of
patients. Using this approach, ≥125 mL of contrast material is
injected at 4 mL/sec. Image acquisition begins after a fixed scan
delay of 20 to 25 seconds. A fixed scan delay is less successful in
patients with poor left ventricular function or another
contraindication to a large contrast volume. In such cases,
delivery of a small timing bolus should, at least in theory, enable
tailoring of contrast administration. The scan delay is the time to
peak arterial enhancement plus 4 seconds, to ensure opacification
of the entire pulmonary tree.
In my experience and that of other researchers,
the use of a timing bolus in pulmonary CTA is time-consuming and
less successful than expected. Enhancement of the pulmonary
arterial circulation is more variable than that of the systemic
arterial circulation, in part as a result of variations in venous
in-flow during respiration.
Hartmann et al
compared individualized with fixed scan delays and found that image
quality was not significantly different. In each case, just over
60% of images were of good quality, 32% to 34% were of moderate
quality, and the rest were of poor quality. The mean contrast
transit time was 10.5 seconds (range 4 to 26 seconds). Contrast
transit time did not correlate significantly with heart rate, blood
pressure, patient size, or cardiac function, further confirming
that pulmonary arterial enhancement is highly variable.
In selecting a contrast agent, it is important to remember that
pulmonary CTA is a first-pass study. It has been suggested that the
higher viscosity of isosmolar contrast medium may result in more
sluggish flow and improved enhancement of the pulmonary arteries.
However, as pulmonary CTA does not involve subsequent acquisitions,
higher retained enhancement offers no advantage.
Two recent studies argue against the superiority of isosmolar
contrast medium in pulmonary CTA. Goodman et al
compared the vascular attenuation achieved when isosmolar and
low-osmolar contrast agents were used in CT venography after
pulmonary CTA. They found that iodixanol was associated with a
modest 7 HU improvement in venous attenuation, but they also found
a 42 HU reduction in arterial attenuation (
<0.05 in both cases).
A separate study by Bedard et al
also showed no significant improvement in pulmonary arterial
enhancement with isosmolar contrast medium, when compared with
low-osmolar contrast medium. Mean pulmonary arterial enhancement
was 257 HU and 252 HU, respectively.
Given the small study size and the complexity of pulmonary
arterial enhancement dynamics, there may have been too few patients
and an insufficient hemodynamic range to demonstrate significant
results. Nonetheless, the role of isosmolar contrast medium in
pulmonary CTA remains uncertain.
One exception would be in triple rule-out studies using a
64-slice CT scanner. This type of study requires optimal
enhancement of the small pulmonary arteries, the coronary arteries,
and the aorta all at once. In such a case, isosmolar contrast
medium is indicated, as it is ideal for opacification of the
coronary arterial tree.
Rapid, recent advances in MDCT technology have created
tremendous opportunities for CT imaging in cardiovascular disease.
The role of CT in pulmonary embolism and aortic disease has long
been recognized. While the exact role of cardiac CT in the
evaluation of coronary artery disease has not been fully defined,
it is quickly being accepted as a worthwhile test for patients with
suspected coronary artery diseases.
ELLIOT K. FISHMAN, MD:
Cardiac CT is as exciting as it gets. At a course a couple of weeks
ago, I asked how many of the 300 people there were doing cardiac CT
and there were probably 10 who said they were. I asked how many
planned on doing it, and it was about 300. How do you get started
in cardiac CT?
ELLA A. KAZEROONI, MD:
Well, I think a lot of people want to do coronary CT, and they're
afraid to get started. As you said, 10 people are doing it, but 300
want to, that's the reality that I've been seeing, too. There are
some review articles in the literature you can read, but I think
what you really need to do is to talk to someone who's done it.
Pick up the phone and talk to somebody, and I think there are
experts around the country who would be happy to help.
Find out what the pitfalls are and get over your fear because,
yes, it's a complex exam. Yes, you have to use beta-blockers, which
I think is a little bit of people's fear, as they have to set up a
nursing protocol to monitor the patients. But it's really not
rocket science. There are standard protocols that they can use,
there are standard viewing tools for evaluating them, there is a
standard nomenclature for evaluating the coronary artery segments,
and there are standard ways of giving the beta-blockers. So, I
think there is a lot of unnecessary anxiety.
CT radiologists aren't very used to giving medications before a
study. So the idea of, "Oh, I'm going to give a beta-blocker and
some-body's blood pressure might drop," may be holding people back.
But it is just a small dose of an oral beta-blocker 1 hour before a
scan or an IV beta-blocker, and having somebody check their heart
rate and blood pressure. We haven't had a complication yet. So
maybe it's an unnecessary fear--you have to take a protocol,
establish it, and start doing it.
JULIA FIELDING, MD:
We've found it useful to work with the cardiologists from the
beginning, and we actually had a cardiac fellow assigned to us.
That did two things: It brought patients down because one of the
federal requirements now is that fellows learn about it; and he
also took care of the drugs. For a lot of patients, we're trying to
make it as joint a service as we can right now to hold it. But is
this a reimbursable exam?
We've found that we are generally getting reimbursed for patients
who have a symptom, for example, chest pain or discomfort, or
people who have a known diagnosis of atherosclerotic coronary
disease already, if they've had a stent or bypass surgery. What
we're not generally getting reimbursed for are the asymptomatic
at-risk individuals. While there is going to be a new CPT code for
coronary CTA, it will still be some time before it is assigned an
RVU value for reimbursement. So, for symptoms or known disease,
most people will bill these cases as a chest CTA code.
Now, you could make the case that a coronary CTA could be the
equivalent of a treadmill test, but with direct visualization of
the coronary arteries. I have heard people say they have gone to
their third-party insurance carriers and made the argument that a
treadmill test is a screening test, yet you're reimbursed for that.
Why aren't you reimbursed for a coronary CTA? It may just be too
early, so we don't have enough data to show exactly what the
indications are for coronary CTA. But to date, the asymptomatic
at-risk people getting screening coronary CTAs have not been
reimbursed. So, we have a fixed price for people who are going to
get one of these to pay up front.
GEOFFREY D. RUBIN, MD:
This is an evolving area, though, because the ACC and the ACR have
been getting together about new CPT codes, which are likely going
to be what's called Category 3 codes, and will not have RVU values
associated with them because the technology at this point probably
doesn't meet criteria from the standpoint of proven outcomes to be
category 1 codes. So, in fact, there's a real likelihood that in
the short term it will get harder to get reimbursed for the test
than easier. It will be up to third-party payers how they want to
reimburse this code, but there will be no value associated with it
from CMS. Right now you can use a thoracic CTA code in many cases
with appropriate ICD-9 codes for chest pain and such, and it makes
sense. But once there's a dedicated coronary CTA code, then you're
going to really have to use that.
I wanted to know also about your triple threat, because I actually
feel that's where radiology can work well because we control the ER
patients right now. In fact, the cardiologists don't even want to
come downstairs to see the ER patients if they possibly can. One of
the big problems has been a high radiation dose to the chest if you
had to do more than one pass. How do you anticipate doing the
triple rule-out with the minimal radiation dose possible?
KAZEROONI: I think if you're willing to keep your contrast
volume on the higher side, with one pass you'd be able to get the
nary CTA, the aorta, and the pulmonary arteries. As long as you
keep the volume of contrast flowing so that when you're scanning,
everything is opacified, I think you can do it.
That would probably do it.
With regard to your comment about the ED, it really depends on
local variation. If you have a busy chest pain center, that's often
run by the cardiologists, and they are the direct input in the
door. The scanners in the hospital are generally owned by the
hospital, not by the radiology department or the cardiology
department; so, in fairness, any practicing physician who has
privileges can have access to any machine they want if they have
the privileges to do so. So, simply that we currently do all the CT
imaging for the ED does-n't necessarily mean that we would always
be doing the ED imaging.
I'd like to just comment on your point about working with
cardiologists and, in particular, to have the cardiac fellow give
the drugs. I think that there are going to be lots of models that
evolve, and people are in a wide variety of practice circumstances
where that may or may not be possible. A point Ella made that's
very important is that, by and large, giving beta blockers and
nitro, which is something else that we do, is a low-morbidity
event. It's much better if a radiologist or whoever is going to do
the procedure gets over the hump of just saying, "I'm going to give
the drugs." Then it's not a problem of having to rely on a crutch
of having somebody come down to help them give the drugs because
it's just simply not that necessary. There are lots of reasons to
work with a cardiologist in this endeavor, but I don't think giving
beta-blockers is really one of them.
However, when the patients come down to us, they are often sick,
and sick enough that I'm not all that thrilled about having them
not monitored by a physician while we're putting them on a scanner.
Are you willing to give the beta-blockers and not be present? That
would be an issue to me, I think, depending on the sickness level
of the patient.
Well, for our patients that come down from any of our intensive
care units, if they're sick enough to be in the CCU, they come down
with what we call our "SWAT team," which involves a nurse,
respiratory tech, and, sometimes, a resident as well. But there's
always a nurse who comes down with them on the SWAT team.
Then you would write your order for your meds and give it.
We write the orders for the meds and our nurses give them. For the
outpatients, we have a number of nurses who work in the radiology
department, and the number of radiology nurses seems to be
increasing every year. The nurses evaluate the patients. They have
a standard checklist where they ask patients about
contraindications for beta-blockers; they check their heart rate
and blood pressure. They give the medication and then check
serially over the next 30 minutes. As soon as their heart rates get
down to the 60-70-ish window, then we take the patients back to the
scanner. We start with a 50 mg dose of oral metoprolol, and then if
the oral hasn't worked, we give an IV dose. But we had a patient
recently with a heart rate of 120, who was already on oral
beta-blockers. Now, for that patient we had to go to the next step,
which as I mentioned was an esmolol drip. That is, I know, a fear
that many radiologists would have a harder time overcoming. But we
went to the ED and got the esmolol protocol. It turned out that
several of the radiology nurses are former ICU and/or ED nurses, so
they were very comfortable with running the drip, and we had no
problem giving it. Giving an oral and/or IV beta-blocker is low
I don't have any problem with oral. I give beta-blockers all the
time for renal biopsies, actually, for hypertension and stuff like
that. So I think that can certainly be done.
Our evolving thoughts on beta-blockers are kind of interesting.
There's a real desire to want to just make it an oral agent and
give it to people and have that be satisfactory. But more and more
I'm coming to the opinion that oral beta-blockers are probably not
worth the time it takes to give them, particularly for outpatients.
Because what we do is we bring the patients in an hour ahead, give
them the beta-blocker, and then they sit there. Usually, their
heart rate does lower. However, the one thing that you can't
anticipate from a resting heart rate measurement is the lability of
their heart rate once you give the contrast, and it will be labile.
So in any setting, but particularly for outpatients, going
primarily to IV and skipping the oral could save you an hour of
having the patient there and might give more reliable beta blockade
that in the end will reverse quicker than giving the oral agent on
top of what might be IV. I think this is going to be an evolving
trend, but I'm coming to the feeling that it's just easier to bring
people in, put them on the table, and give them IV beta-blockers
and skip the oral completely.
I think in the long term probably you're right. We do the same
method. We give 50 mg of metoprolol, hook them up, and get their
heart rates. Most of the patients convert to the mid-60s by 30
minutes, and I would say that in 1 out of 10, or 1 out of 15, we
have to then give the IV. In about 45 minutes, if they haven't
dropped to a good mid-60s, we'll just bring them in the room and
But do you watch their heart rate during the scan?
Yes, I do. First of all, when we test their heart rate, we have
them hold their breath outside, because that, again, could change
things significantly. Remember that even doing non-beta-blockers in
patients with noncontrast coronary scanning, as soon as the machine
makes that noise, the patients get a little apprehensive and their
heart rates increase. That's been shown from cardiac cath as well.
The pre-injection and postinjection heart rates tend to be within a
few heartbeats. So that's worked very well.
I think you're definitely right--it's great that radiologists
work with cardiology. But if we want to develop that area of
cardiac imaging, we need to take total responsibility for the
entire process. The radiologists need to know the contraindications
to beta-blockers. So if you're an asthmatic and you're on inhalers,
that's a contraindication to beta-blockers. So I think it's
something we probably did like the rest of you, you do in-services,
you train the nurses, you train the physicians. But unless you're
willing to do the good and the bad part of the study, it's not
really going to work for us.
I think it's a challenge, but I think 64-slice scanning makes it
much easier. On 16-slices, you really had to get all your ducks in
order to make the cardiac imaging work. On 64-slices, it is pretty
easy. It's pretty easy to get a high-quality data set essentially
every time, and the throughput is really terrific. As Geoff said,
if you're doing IV and not PO, it should make it a lot easier not
to get the timing down perfectly. One thing I wanted to ask in
terms of the medication, is do you use nitroglycerine on your
We may give a sublingual oral nitro before the study. A year ago we
weren't doing that. It's something we've been doing for about the
We do the same, and I was going to mention that the nitrates are
another factor that will raise the heart rate because, in order for
the nitrates to be effective, you need to scan about 5 to 7 minutes
after giving it. Actually, our colleagues in cardiology who did a
study looking at coronary vasodilation every minute after giving
sublingual nitroglycerine, found that 5 to 7 minutes is really the
peak time. So you can time the scan with that intent.
Do you give the pill or do you use the spray?
We have to give the pill because of pharmacy regulations. We are
not allowed to reuse the spray on multiple patients. Obviously,
that would be much better, and my preferred delivery. But we would
have to use a separate bottle for each patient, so we use
sublingual pills instead.
You give the sublingual nitroglycerine and then you give them the
No. If we need IV beta-blockers, we give those until we have the
target heart rate we need. Then we do an unenhanced
calci-um-scoring scan. Once that's completed, we give them the
nitrates and set up the contrast injection in time for our 5- to
7-minute delay after the nitrates are on board and do the CTA.
So you'll just wait 5 to 7 minutes for it to take effect. What is
the risk? I know cardiologists have done that forever when doing
coronary caths. Do you see any issues in terms of changing the
degree of stenosis using nitrates?
If anything, I think it's helpful because the idea is in an area of
thick stenosis, the coronaries should not be capable of
vasodilatation. However, in the more normal zones, it should. So it
should accentuate the difference between the normal and abnormal,
and you definitely can see the difference in the size of the
The cardiologists have been doing this routinely for cardiac caths.
As we make comparisons of CT to cardiac cath, as far as where the
stenosis is or how severe the stenosis is, if the CT is without
nitrates and the cath with nitrates, you're going to be comparing
apples and oranges, so you want to make the cardiac CT as close to
that cardiac cath as possible.
STANLEY GOLDFARB, MD:
Listening to this as an internist, the problem I hear is that
you're assuming an understanding of the patient's condition and the
safety of these types of maneuvers in a given patient. For example,
suppose you have a patient with hypertrophic cardiomyopathy that
was unknown to the patient and the referring physician, and now you
raise the transaortic gradient and the patient has a cardiac
arrest. It can certainly happen in the cardiology suite as well,
but the clinicians would be prepared for that sort of complication.
So, not only do you need to be able to handle the next routine
case, but you also need to be able to handle all the unexpected
We've been thinking about using nitroglycerine and we haven't. I
agree with you, the image quality does improve, but we were pretty
good without it. I think there is no doubt that it's an evolving
The other question I had is for this triple study--there is a
lot of interest in that, obviously. You say that you increase the
volume. What volume would you do for the triple?
Well, if you need to cover everything, you have to think of the
aorta, the coronaries, and the pulmonary arteries. So you have to
go up to what the highest one was, and, typically, that's for the
pulmonary arteries, where we usually use 125 to 135 mL. So we go up
to that number for the full study.
But what about the issue with the right side of the heart and the
saline push and everything else?
I only partially buy into doing the saline push for the right side
of the heart. Clearing the contrast out of the right heart may give
me a little more information about the right coronary artery. For
gated cardiac exams done for other reasons, such as pericardial
disease, masse and mapping for ablations, we see the right coronary
artery very well. So, I'm not sure I need to clear the contrast out
of the right side of the heart, and I haven't found it to really be
Our injection protocols for the coronaries are 80 mL of isosmolar
contrast at 4 mL/sec and a 40-mL saline push. The studies all use a
test bolus, but they are coming out to about a 28- to 30-second
delay. We've always found that the saline push is very good for
looking at the right coronary arteries because you really do clear
the right side of the heart out. When we didn't do that, we often
have a lot of turbulent flow, which creates an artifact, and you
can often not get a good visualization of the right coronary. The
very proximal part, you can, but depending on the coronary set, you
have so much artifact over it. That would be the only concern I
have for that triple study.
I agree. It also buys you efficiency with the contrast, and it's
well established that when you give a saline push, you will get
more enhancement per gram of iodine delivered because you've pushed
it out of the central and peripheral veins and gotten it into the
pulmonary and systemic arterial circulation. Right now, we
certainly prefer to use the saline push, too. On this topic, you
had mentioned two approaches to doing the pulmonary CTA, the fixed
delay and the test bolus, and had mentioned that the test bolus
might be a little less reliable. But I would consider a third
approach, which is the one that we actually use, which is direct
Yes, we use that too.
The reason it is separate from a test bolus is because the
physiologic impact of a 15- or 20- mL test bolus is totally
different from running in your primary bolus. I think there's much
better association between what you measure in a pulmonary artery
when you're actually measuring your primary bolus than there is
when you run that little test bolus.
We've tried that method for pulmonary artery CTs as well, and we've
found that the time it took to move from knowing where we were
measuring enhancement to getting to where the scan started was too
great, and that it was affecting our pulmonary arterial enhancement
too much. Pulmonary arterial enhancement doesn't appear to be an
on/off phenomenon, like the aortic/arterial side of the
circulation. Especially when the patient's breathing changes, which
rushes nonenhanced blood from the legs and abdomen into the IVC and
right atrium, diluting things. The contrast really wasn't still on
its up-slope; it was going to go down and come back up again. So we
found that really didn't work too much.
Geoff, where are you triggering it? At what number are you
We trigger off the right ventricle, first off. So we're pretty
close to where we start because we scan our pulmonaries from bottom
to top. But also, we take a substantial delay after the contrast
arrives, so once the contrast arrives, we wait 10 to 15 seconds
before we actually start the scans. We don't have an enhancement
threshold; it's visual. When you do these cases, they typically go
from 40 to 150 or 200 HU. It's not like a test bolus where you have
subtle changes. It jumps up very quickly so the actual number isn't
so important. You know when the contrast has arrived. But the point
is that not only will poor left ventricular function alter your
timing, but so will SVC stenoses because of in-dwelling catheters.
There are lots of potential problems that can conspire to hurt you.
Even if somebody has had chronic PE and has tricuspid regurgitation
and is blowing a lot of contrast in their abdomen, all of these
variables can impact it. The Valsalva maneuver that a person
performs when they take a deep breath can cause the types of
problems for arterial opacification that you're alluding to.
So I think the one thing we can do on the patient side to help
avoid that is to train them not to Valsalva when they take their
breath-hold for the scan because when they Valsalva, what that ends
up doing is pushing a lot of unopacified blood.
So what volume do you use for your CTA studies?
We use 120 mL.
I think the one study of angiography where people fail the most, or
feel they fail the most, is pulmonary emboli studies. The comment
people always make is, "I do the studies the same exact way. One is
perfect, the next one is absolutely terrible," and there's nothing
any different. So your recommendation would be to use a visual
target as it comes in the right ventricle. How many seconds are you
Fifteen seconds, then bottom up.
Right. You know, whether it's bottom up or top down; I don't have a
strong opinion about that anymore. It used to make a difference
with single-row CT when it was such a long breath-hold. We do it by
convention mostly at this point.
The big thing for us, also, is that you want to move the machine as
little as possible because often you have to move to the top, but
that'll add another 5 seconds.
It takes a little longer. I will say that as you get to faster
scans, if the patient has a period of poor opacification because of
a Valsalva--let's say it's 2 seconds--on a 16-row that 2 seconds
represents a lot more anatomic coverage than it does on a 4- or an
8-row scanner; on a 64-row, it's even more. So, with the faster
scanners, it's more of an all or none--good study or bad study. On
a 4-row, you had a little band that was unopacified. Now, half the
chest is potentially unopacified. So that's where the risk comes
But I think that's when people speak about how they overread PA
studies because they look and everything's white, white, white, and
there's a couple of slices where things aren't so white. It's not
like a thrombus there. It's just the relative lack of good
enhancement. So that's probably very critical.
In terms of postprocessing, one of the key things I would agree
with is your comment that right now the best way of doing a
coronary artery study has not yet been determined. Whether it's
axials or curved coronals, MIP and volumes, and we use a little bit
Right. We do, too.
So, do you have a recommendation of how you would tell someone how
to interpret a routine coronary study?
The first thing I do is to just quickly scroll through the
noncontrast axial images to get a sense of the calcium burden
because that's going to impact whether I may even be able to
interpret the coronary CTA. If you've got a calcium score >1000,
I may very well not be able to interpret the coronary CTA in a lot
of areas, so it gives me a road map.
If someone has a very high calcium score, and it's incredible
calcification, are you still doing the CTA?
Well, since calcium scoring has to be processed after the fact,
after the scan has been acquired, we don't say, "No, we won't do
the coronary CTA."
People have talked about if a calcium score is over a certain
number, if you see a certain amount of calcification, you shouldn't
do the coronary CTA. I haven't come to the point where I've said I
won't do it if there's a lot of calcification.
But it gives me a sense of what I'm up against when I go to look
at the coronary CTA. Then I start with curve reformats, and I like
to move the volume around and interact with the data set and try to
draw out the small coronary arteries using the curve reformats. I
also have a lumen view running so I can stretch out the artery and
rotate it and look for whether the stenosis is excentric or
circumferential. The other thing I like to have is the cross
section, to be able to see the cross section there, so I've got the
length of it and I have the cross section together, and I like to
have both to combine. All the fancy shaded-surface displays and 3D
looks are pretty. They look nice in talks; the cardiologists and
their patients like them. But I don't think that's really where the
diagnosis is. That's just, "Hey, we've got a beautiful study here
today." But I think needing to have both the cross section and the
long view of the artery is important.
When you interpret the study, what do you say? Your final report
will be, "Oh, just a normal case." But do you measure stenosis? Do
you try to quantify above 50%? Under 50%? How do you report it?
We try to report it the way our cardiology colleagues do, to use a
similar style as they would in their cardiac cath report.
We measure diameter, percent diameter stenosis. It's not the
area of stenosis, but in our facility, most cardiologists report
percent diameter stenosis. So that involves either manually or
using some software to do it for you, determining what the diameter
of the vessel is above the stenosis, below the stenosis, and in the
stenosis. By knowing above and below, you estimate what it should
be at the stenosis, and then from there you determine the percent
diameter stenosis. We report anything >30% stenosis; anything
<30% stenosis, we say that there's plaque there, but we don't
One of the unique aspects of reading these cases is dealing with
the temporal domain. It's absolutely critical to move through the
different phases to find the right view for the different
coronaries. So our PACS, which is fairly rigid in that you can
either show all the images sequentially or by table position,
doesn't work for that. You might think that looking by table
position would be a good way because you look at 10 phases at one
position, and then you go to the next, but it's not a very logical
way to navigate.
So we use a workstation exclusively for these cases that allows
us to bring up a volumetric view. It might be volume rendering or
multiplanar reformation. Then from any view we can click through
the 10 phases, and I think that gives the optimum access to what
can be up to 4000 images. I probably have a pretty antiquated
approach. I feel most comfortable looking at cross sections, and I
start out with axials and find the most motion-free views for each
vessel and work my way down. I turn the planes so that I'm always
perpendicular to the vessel, and it's really a manual turning. I
occasionally go parallel to the vessel. Diagnosing intrinsic
stenosis and myocardial bridging, which comes up a fair bit in
these patients, is really best seen on those kinds of views. I love
to create the volume rendering and they're like a reward at the end
of the case. But rarely are they particularly useful
diagnostically, except perhaps for aberrant coronaries. That's when
you really like to have the volume renderings.
What about MIP?
I find MIP almost worthless, unfortunately. The problem is that you
start dilating out a little bit, and, even on a thin-slab MIP, any
calcium wipes you out, and the myocardium starts to get in the way.
I just really believe that cross sections are the way to go. If I
had a convenient way to do curves on my workstation in real time, I
would probably use those. If I could create one curve and then have
it interpolate across all 10 phases and click through those 10
phases, that would be a pretty swift way to go.
I think what you're speaking to is that the software to evaluate
the cardiac CTs is much more primitive than the high-tech scanners
that generate them. We generate this very powerful data set, but
the software tools have really lagged behind a way to efficiently
evaluate this data.
You have to deal with what you have. I'd also like to note that
it's very unusual for me to measure a stenosis just because I don't
think I can believe the measurements. If you look at the number of
pixels spread across the coronary, than the best I can hope to do
is say whether it's <50% or >50%. If I think it's high grade,
≥90%, I will say that. If it's occluded, I will say that. But I try
not to measure.
What you said is very true. We're doing cardiac imaging, which is
the most demanding vascular imaging, but we're doing that with
tools that were developed for looking at the abdominal aorta.
I think one of the things we will all see this year will be
dedicated cardiac packages that are really optimized cardiac
visualizations, whether it's spider views or whatever else. A
number of vendors are working on these. That will be a major factor
in cardiac imaging.
I think it's a very exciting area. Some of the things you
mentioned, such as the functional valve imaging, are just going to
continue to grow in that area.
I would like to add one point. I've found that the greatest impact
on getting our referrals was giving internal medicine grand rounds.
Immediately after the grand rounds, we saw our referrals bump up
substantially, on the order of 4 to 5 cases per week. You can reach
a lot of key referrers through that one presentation.
You said you did calcium scoring first, which is similar to us, and
we have the coronary CTA. Do you examine cardiac function on every
case? Do you provide ejection fraction?
We want to and we will, once the software is reliable for it. We
have had some problems with the nuts and bolts of the DICOM data
coming from the CT scanner meshing with the software products that
we have to do it. But then we will do that.
We have one of our faculty doing cardiac function on nearly
everybody, and we're doing valve analysis on every patient. We're
at the point now where we can compete in this arena, so you have to
do more than they ask for. It reminds me of that TV commercial
where they show a kid mowing lawns and he always gave customers
more than their money's worth--he even manicured the lawns and
That's what we need to do. Even when they just ask us to report
on the coronaries, we're giving them the AVIs and looking at the
aortic valve. That's generated a lot of business, for example, from
the cardiothoracic surgeons, looking at ascending aortas, looking
at valve motion, looking at the coronaries at the same time, and
trying to bypass other exams. It's critical for us, if we want to
develop that area, to do more than they ask for, show what you can
do do--as we've done everywhere in radiology. It is definitely time
consuming. The biggest challenge to radiology is for people to be
willing to put the time in to do it, from being able to give
medication safely, to learning all the postprocessing steps and
providing the information.