Josef Machac, MD, The Mount Sinai School of Medicine of
New York University, New York, NY
Myocardial PET perfusion imaging has been used clinically for
over a decade. During this time, both SPECT and PET imaging
technology and expertise have improved. However, significant
problems with cardiac SPECT accuracy remain, particularly in
women and the increasing proportion of obese patients. PET
perfusion imaging offers high specificity in these and other
difficult patients. Improvements in modern PET gamma cameras
allow gating of Rb-82 images, offering the possibility of wall
motion acquisition, routinely performed with cardiac SPECT
imaging. In addition, there is a need for implementation of
practical blood flow quantification to provide valuable
information not available by other means, including quality
control for pharmacological stress imaging, the evaluation of
small vessel disease and endothelial dysfunction, detection of
"balanced ischemia," and the detection of coronary
This article discusses how these changes offer an opportunity
to apply the full power of cardiac PET imaging. The shift of the
clinical paradigm from disease detection to prognostic risk
stratification dictate an update in clinical cardiac PET imaging
outcome studies. These will serve to define the role of
myocardial PET perfusion imaging in the next decade.
Josef Machac, MD
Myocardial PET Perfusion Imaging: The Next
Clinical positron emission tomography (PET) imaging is
experiencing rapid expansion due to the recognition by the
clinical community of its value in decision making in oncology.
Cardiac PET imaging had an early recognition of its value in
myocardial perfusion and viability imaging.
However, its limitation to a few PET centers has retarded its
clinical development. The recent rapid growth of PET imaging
systems offers a new opportunity for cardiac PET imaging. It is
also a challenge for clinical cardiac imaging to take a quantum
leap to a new level of functional imaging capability.
Principles of PET Imaging
PET imaging utilizes a class of radionuclide tracers that
decay with positron emission. This leads to the production of two
511 kev photons travelling in opposite directions. A PET camera
is able to detect the two gamma rays in coincidence. The last
decade has seen the development of high-performance dedicated PET
cameras featuring high sensitivity, high resolution, high speed,
and larger fields of view.
For cardiac imaging, high resolution is not currently a major
concern. Myocardial contraction results in smearing of the image,
unless one performs ECG gating, where high count sensitivity,
rather than resolution, is critical for good image quality.
Moreover, for clinical myocardial imaging, only an appreciable
mass of myocardium that is hypoperfused or dysfunctional is of
diagnostic or prognostic significance.
Image uniformity is by far the most important property of
cardiac PET perfusion imaging. Figure 1 shows SPECT images of a
cardiac phantom, showing progressive attenuation from the apex
toward the base. Experienced readers are familiar with this
pattern and learn to distinguish it from real defects.
Sometimes, this is difficult to do. On the other hand, a real
perfusion defect can be hidden within an area of apparent
attenuation. PET imaging corrects this problem.
The effectiveness of cardiac SPECT imaging has been greatly
augmented by ECG gating. Gating provides added information about
global and regional left ventricular and right ventricular
function, enhancing its prognostic value,
and helps to differentiate real defects from attenuation
However, further improvement is needed.
Among the tracers used for PET perfusion imaging, N-13 ammonia
allows high quality static and gated imaging,
as well as myocardial blood flow quantification at rest and
However, the requirement for an on-site cyclotron for its
production limit its clinical use. Even in PET centers with a
cyclotron, the current lack of reimbursement imposes additional
O-15 water allows reliable quantification of blood flow.
However, poor image quality and the requirement for on-site
cyclotron production just before each use preclude its routine
Rubidium-82 (Rb-82) is a potassium analogue that, like
thallium-201, is extracted by all living cells. It is produced
from a commercially available, FDA-approved strontium-82
containing generator, which must be replenished 13 times a year.
The short half-life of Rb-82 allows repeated acquisitions every
10 minutes. Its round-the-clock availability makes rubidium-82
the most practical PET perfusion imaging agent. The short
half-life of Rb-82, however, imposes a limit on the available
imaging time for each injection, and a limit on the obtainable
image counts. It requires a high-sensitivity, high-speed PET
scanner, which is able to handle the high injected dose (50 to 60
Clinical myocardial perfusion PET imaging with Rb-82 consists
of positioning, transmission imaging, and resting and stress
imaging. The resting acquisition takes 6 minutes. It can be
performed with ECG gating. One obtains an 8-frame gated
myocardial wall motion/perfusion study for wall motion evaluation
and quantification similar to gated SPECT imaging. The resting
imaging is followed by a 10-minute transmission imaging,
performed with a rod source of activity (Germanium-68), housed
inside the scanner cabinet, which circles the body. The
transmission images provide information for attenuation
correction, which is a crucial component of cardiac PET
The final step is pharmacological stress imaging. The most
common stressors are dipyridamole or adenosine, which increase
blood flow three- to four-fold in normal regions.
Dobutamine or arbutamine can be used in patients with asthma who
cannot tolerate dipyridamole or adenosine. At peak stress level,
another intravenous Rb-82 infusion is delivered, followed by
stress imaging for 6 minutes. The stress images can also be
gated, with resultant wall motion information during stress.
Other stressors can be used, including mental stress, smoking,
handgrip, and the ice-pressor test for special applications.
Figure 2 presents a normal PET perfusion study and the end
diastolic and end-systolic frames from the gated resting study.
Figure 3 shows PET perfusion images of a patient with multiple
perfusion abnormalities at stress and rest, LV dilatation, and
regional and global LV dysfunction.
With a reported high sensitivity (92% to 95%) and specificity
(95%) of disease detection by PET myocardial perfusion imaging,
one can argue that all patients should have myocardial perfusion
imaging studies with PET, if available. Analysis by Patterson et
revealed that, despite higher cost per study, PET imaging is
cost-effective by decreasing the utilization of more costly
angiography and intervention procedures. This was supported by
utilization outcome studies by Merhige et al.
On the other hand, this conclusion was challenged by analysis
using a different model and assumptions.
A conservative approach selects patients with a high
likelihood that PET imaging will yield added value compared with
SPECT or other noninvasive tests. This includes obese individuals
and women with large breasts, in whom SPECT imaging is less
Many patients with end-stage renal and liver disease have edema,
ascites, and high elevated diaphragms, sometimes with pericardial
effusions, which may lead to nonuniform attenuation
abnormalities. Patients with equivocal or conflicting test
results also benefit from PET imaging.
Figure 4 illustrates a stress SPECT Tc-99m sestamibi study
performed for pre-operative risk stratification in a 370-lb man
with an abnormal ECG and multiple risk factors for CAD. It shows
a moderate to severe inferoposterior defect, and a possible
moderate anterobasal defect that were not corrected the by SPECT
attenuation correction. They were considered to be likely due to
attenuation artifact, but real disease could not be dismissed.
The Rb-82 PET study in the same patient showed no evidence of the
inferoposterior or anterobasal defects, at rest or stress. This
patient would have been incorrectly assigned to a high-risk
category, but instead belonged to a low-risk category. The
problem of accurate diagnostic and prognostic imaging in
moderately and markedly obese individuals is likely to grow, in
view of the increasing average weight in the developed world each
Diagnostic accuracy of noninvasive imaging in women has been
problematic. The added value of PET imaging in women compared
with SPECT imaging has been documented.
Figure 5 presents an example of a 52-year-old woman with evidence
of anterior wall ischemia on a SPECT Tc-99m sestamibi study and a
normal Rb-82 PET study, suggesting an attenuation artifact.
Flow Reserve Measurement
PET imaging offers the potential for quantification of
myocardial blood flow at rest and stress with Rb-82. Without flow
quantification, one may underestimate or even miss extensive or
diffuse disease, and the so-called "balanced ischemia."
Alternately, one may fail to produce sufficient vasodilation
during stress, leading to a false-negative study.
Formal quantification of blood flow requires a multi-frame
dynamic acquisition. Myocardial and blood-pool activity curves
must be generated and corrected for decay, the partial volume
effect, tissue cross-talk, and dead-time. A compartmental model
is used to solve for multiple unknowns, including blood flow.
This approach is difficult for routine use. Simpler
alternatives, even if less rigorous, can still be useful. The
simplest one uses the ratio of Rb-82 uptake during stress and
Their ratio reflects the true flow reserve, but ignores the
effects of cardiac output increase during stress on Rb-82 blood
pool activity, and the decreasing extraction fraction of Rb-82 as
coronary flow increases. Thus, the uptake ratio underestimates
the true flow reserve fraction. Nevertheless, the Rb-82 uptake
ratio has been used successfully as an index of the flow response
to stress and to detect the coronary steal syndrome.
A compromise alternative uses modified equations for the
trapping of microspheres in tissues.
It corrects myocardial Rb-82 uptake by the summed blood pool
activity and by the relation between flow and extraction fraction
obtained from animal studies.
A global failure to achieve a normal flow reserve ratio
greater than 2.0 to 2.5
suggests a failure in vasodilator efficacy; small-vessel disease
in such situations as diabetes, hypertension, or endothelial
dysfunction; or extensive epicardial disease, the so-called
"balanced ischemia." Studies show improvement in flow reserve
with interventions. Endothelial dysfunction has been studied
noninvasively using the cold-pressor test.
Physiological interventions can easily be performed with Rb-82
A 70-year-old man with risk factors for CAD and a positive
exercise ECG stress test was evaluated in our laboratory for
severity of CAD. The PET images showed only a small apical
defect, which improved at rest (Figure 6), indicating only mild
disease. However, the measured flow reserve was very low (1.2).
On the suspicion that the low flow reserve was due to inadequate
stress stimulus, small vessel disease, or "balanced ischemia,"
the patient underwent angiography, which showed three-vessel
disease. This extent of disease would have been missed, since
cavity dilatation or a positive ECG response with dipyridamole
that make one suspicious were absent.
The accelerated atherosclerosis following heart
transplantation tends to be diffuse and involves both epicardial
and small vessels diffusely.
Such patients are ideal candidates for follow-up with flow
reserve quantification by PET.
Reliance on regional heterogeneity alone may underestimate the
extent and severity of disease.
Quantification of blood flow also can reveal the presence of
collaterals to diseased regions. Myocardium supplied by
significantly diseased arteries does not increase blood flow to
the same degree as normal regions. In multivessel disease and the
presence of collaterals, blood flow with stress may actually
decrease, demonstrating the "coronary steal syndrome," which can
be detected by quantification of regional blood flow.
The PET imaging study in Figure 7 is of a man who presented with
recurrent chest pains showed severe extensive periapical and
lateral defects, with essentially normal resting perfusion. While
the flow reserve in the septum is nearly normal, the lateral wall
showed a 30% fall in flow with stress, an example of coronary
steal. On angiography, the patient had three-vessel disease, with
Routine rest and stress PET imaging with Rb-82 provides useful
relevant information on myocardial viability, similar to that of
Tl-201 imaging and Tc-99m sestamibi SPECT imaging. Combined with
wall motion information, the PET myocardial perfusion study can
be expected to answer the clinical question in most patients.
There are, nevertheless, regions with demonstrable viability that
show deficient uptake of Rb-82, an example of hibernating
Patients with poor LV function are best performed with combined
rest and stress Rb-82 imaging and F-18 FDG imaging. HCFA has
recently endorsed reimbursement for FDG PET imaging for
In spite of enhanced diagnostic and prognostic capabilities
with the use of gated SPECT, significant limitations to SPECT
imaging still exist. Early clinical experience with Rb-82 PET
perfusion imaging has demonstrated a consistently high accuracy
for CAD detection. This makes PET perfusion imaging a compelling
first choice in an identifiable subgroup of the clinical
population, those who are difficult to diagnose by other means.
The implementation of routine gating with PET myocardial
perfusion imaging and the development of practical flow reserve
quantification promise to augment the already considerable added
value of myocardial PET imaging.
The current proliferation of dedicated PET cameras is an
opportunity and a challenge to make PET more widely available for
cardiac imaging. In PET centers overburdened with oncology work,
the expansion of PET services to cardiac imaging could provide a
reason for the acquisition of an additional PET camera. Centers
that are hesitant to acquire a dedicated PET scanner because of
an insufficient number of anticipated oncology imaging studies
could have a sufficient number of cardiac studies (2 to 3 per
day) to pay for a rubidium-82 generator and contribute to the PET
camera overhead. The same reasoning could be applied to mobile
At the same time, important challenges lie ahead for cardiac
myocardial PET imaging. Most of validation studies of Rb-82 PET
perfusion imaging were conducted 10 years ago. Since both SPECT
and PET imaging methods have continued to evolve, these studies
need to be updated.
Outcome studies have amply documented the added value of SPECT
radionuclide imaging in prognostic stratification. The literature
on PET prognostic risk stratification in areas other than in
viability imaging largely still remains to be written. Given the
robustness of PET imaging in assessing the severity of CAD, it is
expected that PET imaging will meet this challenge successfully.
In spite of the impressive past accomplishments of cardiac
radionuclide imaging, the pressures for continued improvements in
performance is attested to by current efforts being made with
alternate imaging modalities.