Introduction

Positron emission tomography (PET) myocardial perfusion imaging (MPI) has become widely available and has demonstrated significant value in assessing known or suspected coronary artery disease (CAD) and determining prognosis related to cardiac events, including death and myocardial infarction.1,2 Previous meta-analyses have demonstrated the high sensitivity and diagnostic accuracy of PET MPI for the detection of significant angiographic coronary artery stenosis and have also demonstrated that stress PET was superior to that of both SPECT and stress echocardiography.3,4 Due to the widespread availability of Rubidium-82 on-site generator production and short physical half-life of 82Rb, PET MPI is typically performed using the vasodilator agents, regadenoson, dipyridamole, and adenosine with the patient in the scanner and connected to the 82Rb generator. 13N-ammonia, with its longer physical half-life, is compatible with conventional exercise protocols as the patient does not have to be physically in the scanner during injection. PET studies from institutions with an on-site cyclotron have demonstrated the feasibility of exercise PET using 13N-ammonia and have shown that a treadmill 13N-ammonia PET produced larger perfusion defects than vasodilator PET.5,6 Routine attenuation correction, high spatial and contrast resolution, as well as high temporal resolution are several technical advantages of stress PET that account for its improved image and diagnostic quality compared to SPECT. Therefore, in patients who can exercise, 13N-ammonia PET may be superior to both vasodilator PET and exercise SPECT.

Despite the promising data on exercise 13N-ammonia PET MPI, the limited availability of on-site cyclotrons restricts widespread adoption of the technique. To date, there is no study that has shown the feasibility of performing exercise MPI using 13N-ammonia produced at an off-site cyclotron. In this study, we address that feasibility.

Methods

Study Population

Between February 2018 and October 2019, 4624 patients presented to our laboratory for myocardial perfusion imaging. Of those, 3126 were referred for SPECT. Of those, 2029 were pharmacologic stress and 1097 were performed following treadmill exercise. One thousand four hundred ninety-eight patients were referred for PET, 75 of whom received 13N-ammonia while 1423 received 82Rb. Of the 75 13N-ammonia studies, 33 were performed during treadmill exercise and 42 received regadenoson. Selection for treadmill stress was determined by the ordering provider. If the patient met insurance coverage criteria for PET, treadmill 13N-ammonia was selected as the protocol of choice.

Off-Site Cyclotron Production of 13N-Ammonia

All resting and treadmill exercise stress doses of 13N-ammonia were prepared at an off-site cyclotron run by Wisconsin Medical Cyclotron (West Allis, WI) approximately 7 miles (Figure 1) from our institution. Due to the 10-minute physical half-life of 13N-ammonia, there are several considerations that are necessary to ensure that the drug arrives at our facility on-time for resting and stress imaging at the appropriate doses. Orders for 13N-ammonia are typically transmitted to the off-site cyclotron facility no less than 24 hours before calibration time and are placed into a production planning tool which back-calculates the timing of critical manufacturing events from the estimated time of injection. The planning tool provides a schedule for the start and end of cyclotron target irradiation, the time of end of synthesis (EOS), latest time of departure from the cyclotron facility, and arrival time at the imaging center’s hot lab. Three personnel are required to manufacture, dispense, and deliver unit doses of 13N-ammonia. A production technician will typically prepare the manufacturing systems, materials, and batch documentation about 30 minutes before start of cyclotron irradiation. Following irradiation and purification, 6.0 mL of 13N-ammonia in a sterile vial is available for dispensing. Quality control is routinely performed by the staff of the cyclotron facility. A pharmacist dispenses a unit dose (up to 30 mCi at time of calibration) into a syringe and a pharmacy technician performs packaging and delivery functions, both of which require approximately 10 minutes, while the production technician readies for a subsequent sub-batch of 13N-ammonia. The delivery driver must consider local traffic and plan for an approximately 20-minute car trip, depending on the time of day and weather conditions. Once the driver is en route, staging begins at our institution with the patient on the treadmill ready to begin exercise at the pre-determined time allotting for an average of 6-10 minutes of graded exercise. The Wisconsin Medical Cyclotron (WMC) can repeat the entire process for up to 8 doses each, 60 minutes apart.

Figure 1
figure 1

The Wisconsin Medical Cyclotron (left) delivery person travels approximately 7 miles from the off-site cyclotron facility to our medical center (right)

Full size image

Exercise 13N-ammonia Acquisition Protocol

All studies were performed on a dedicated cardiac PET scanner (MiE, Elk Grove Village, IL) using a rest-stress protocol (Figure 2). The patient is initially placed in the scanner and a positioning scout transmission image is acquired for 60 seconds using germanium-68 rod sources. A laser-guided mark is made on the patient’s chest to document positioning in the scanner. Immediately following the positioning scan, the patient receives a 10 megabecquerel per kilogram (MBq⋅kg) dose of 13N-ammonia up to a maximum dose of 30 mCi. After a five-minute delay post-injection, a 16-minute gated emission scan is acquired which is immediately followed by a 75Kct transmission scan. A 60-minute decay period follows the initial acquisition. Following the decay period, and in coordination with the WMC delivery driver, the patient begins graded, symptom-limited treadmill exercise according to a standard Bruce protocol on a treadmill located in a room adjacent to, and approximately 25 feet from, the scanner. At peak exercise, the patient is injected with a 10 MBq⋅kg (up to a maximum dose of 30 mCi) of 13N-ammonia. Exercise is continued for approximately one additional minute to maximize tracer uptake. The patient is then transferred to the scanner for positioning. There is an approximate 5-minute delay between the peak exercise injection and the stress acquisition to allow for correct positioning and recovery of the patient to reduce motion artifact secondary to breathing. Subsequently, emission and transmission images are acquired as described above.

Figure 2
figure 2

Typical protocol for rest/exercise PET myocardial perfusion imaging using an off-site cyclotron for production of 13N-ammonia

Full size image

Image Processing and Interpretation

List mode emission data were reconstructed into 16-frame gated data sets using 4DM post-processing software (INVIA Medical Solutions, Ann Arbor, MI). Regional myocardial perfusion was assessed by automated quantitative and semi-quantitative visual analysis using the 17-segment AHA model. Standard summed stress, rest, and difference scores were calculated by the software, modified by the interpreting physician as necessary, and recorded. Left ventricular ejection fractions at rest and with stress were calculated by the software. When necessary, operator adjustment of the mitral valve plane was performed. Additionally, the transient ischemic dilatation ratio was recorded. Overall image quality was rated as good, adequate (containing some flaw but interpretable), or inadequate where interpretation was severely compromised.

Statistical Analysis

Continuous variables are presented as the mean ± SD and categorical data as the number of patients with a variable and corresponding percentages of the study population or sub-population.

Results

Baseline Characteristics

Thirty-three patients underwent treadmill exercise (GXT) 13N-ammonia PET MPI from February 2018 to October 2019. Mean age was 63.4 ± 12.3 years and 79% were male. The mean BMI was 33.7 ± 6.9 kg⋅m2. Based on the history obtained by the cardiologist that ordered the study, 88% (29/33) patients were reported to have some degree of CAD based on either prior cardiac catheterization, CT-coronary angiography, or CT coronary artery calcium scoring. Of those 29 patients, 17 were known to have had prior intervention with PCI or coronary artery bypass graft (CABG) (51% of the total study population). Hypertension was diagnosed in 26/33 (79%) patients and diabetes mellitus was present in 9/33 (27%) patients. Twelve patients (36%) were either current or former smokers and 16 patients had a documented family history of heart disease (48%).

Treadmill Exercise

All 33 patients were able to perform graded treadmill exercise according to a standard Bruce protocol. Mean resting heart rate was 73.3 ± 14.1 and resting blood pressure was 132 ± 18.6/74 ± 7.8. Twenty-nine patients (87.8%) were able to achieve at least 85% of their age-predicted maximum heart rate prior to injection of 13N-ammonia. The mean heart rate at peak exertion was 143.2 ± 17.9, and the mean blood pressure at peak exertion was 180 ± 19.9/75 ± 6.8. The average exercise time was 7.6 ± 1.8 minutes.

Dosimetry, Image Quality, Myocardial Perfusion, and LV Function

The 13N-ammonia produced off-site was available in 100% of planned cases. The mean resting dose was 29.2 ± 5.7 mCi, and the mean stress dose was 30.8 ± 7.3 mCi.

Figure 3 shows examples of normal and abnormal rest/exercise studies with 13N-ammonia. Figure 4 compares a patient’s prior sestamibi SPECT with the current 13N-ammonia images.

Figure 3
figure 3

Representative 13N-ammonia myocardial perfusion images from exercise (top rows) and rest (bottom rows). Panel A shows normal exercise and rest myocardial perfusion. Panel B demonstrates a small, fixed inferoapical perfusion abnormality consistent with small inferoapical infarct

Full size image
Figure 4
figure 4

(Panel A) 13N-ammonia PET MPI demonstrates reversible apical defect; (Panel B) demonstrates treadmill SPECT from the same patient 4 years earlier

Full size image

Image quality was rated as good in 29 patients, adequate in 3, and inadequate in 1 patient. The inadequate study was due to patchy myocardial uptake that was a function of suboptimal counts and motion. Figure 5 shows examples of differences in exercise 13N-ammonia image quality.

Figure 5
figure 5

Examples of Image Quality: Panel A shows an example of a Good quality image. Panel B shows an example of an Adequate quality image. Panel C demonstrates an Inadequate image

Full size image

The mean summed stress score was 4.5 ± 5.7, summed rest score was 2.7 ± 5.0, and summed difference score was 2.1 ± 3.6. The TID ratio was 1.0 ± 0.1.

Resting and post-exercise left ventricular ejection fractions were 64.1 ± 11.1% and 66.6 ± 13.3%.

Discussion

Our study demonstrates the feasibility of performing exercise PET myocardial perfusion imaging with 13N-ammonia produced at an off-site cyclotron. All 13N-ammonia doses were received on schedule. All patients performed symptom-limited treadmill exercise with 88% of patients attaining or exceeding 85% of the age-predicted maximum heart rate. Exercise imaging was completed in all patients. Other centers have previously demonstrated the feasibility of performing treadmill 13N-ammonia PET but have had access to on-site cyclotron production of unit-dose 13N-ammonia.5,6 These results challenge the commonly held notion that all PET MPI must be performed with vasodilator stress imaging, relegating all exercise MPI to SPECT imaging. In addition, there is now a small cyclotron available for production of unit-dose 13N-ammonia that can be housed in a nuclear cardiology laboratory or in a department of nuclear medicine. To date, one center has reported the ability to deliver high quality 13N-ammonia PET MPI with the Ionetix ION-12SC (San Francisco, CA).7

Image quality was considered good in all but three patients. In two patients, there was excessive subdiaphragmatic activity, but images were still considered interpretable. By comparison, we found a 10% rate of excessive subdiaphragmatic activity in 42 patients who underwent vasodilator 13N-ammonia PET MPI during the same time period. Liver and/or gut activity may be encountered with 13N-ammonia imaging—its appearance is unpredictable.8 Lung uptake is usually minimal but may be increased in smokers, those with chronic pulmonary disease and in patients with depressed left ventricular function.9 One patient in our cohort had poor image quality considered to be related to lung uptake, motion, and suboptimal counts, the latter unrelated to the 13N-ammonia dose.

Exercise vs. Vasodilator MPI

Current ASNC guidelines for myocardial perfusion imaging recognized exercise as the preferred stressor for those capable of exercising.10,11 The addition of exercise and ECG variables contribute to the diagnostic and prognostic content of the examination. Quantitative scoring such as the Duke treadmill score is a powerful predictor of subsequent outcome that is lost with vasodilator imaging. Additionally, Chow et al. demonstrated larger perfusion defects on exercise PET studies compared with vasodilator-induced defects in the same patients.6,12 PET MPI is also considered better than SPECT MPI in obese patients and Aggarwal et al. have demonstrated the feasibility and accuracy of exercise treadmill 13N-ammonia PET MPI in obese patients.5,13,14

Since patients are on the treadmill and not in the scanner at the time of the injection of 13N-ammonia, it is not possible to measure myocardial blood flow during exercise PET. Myocardial blood flows have been shown to be of diagnostic and prognostic value in the assessment of suspected coronary artery disease. Measurement of myocardial blood flows and flow reserves may help in detection of endothelial dysfunction, microvascular disease, and multivessel CAD and have been shown to be prognostically important in patients with and without CAD.15,16,17,18 To our knowledge, there is no literature directly comparing the relative diagnostic and prognostic powers of exercise PET results to pharmacological PET with myocardial blood flow results. In our institution, the choice of exercise or pharmacologic stress is determined by the ordering cardiologist based on patient-specific factors. In certain patients, the choice would be obvious, such as those with chest pain and normal coronary arteries or those with MINOCA in which myocardial blood flows and flow reserves would be the primary focus of the examination. Further studies are needed to determine other scenarios in which it is more appropriate to sacrifice the prognostic information contained in an exercise study in order to obtain non-invasive assessments of myocardial blood flow or vice versa.

Radiation Exposure

Current practice guidelines continue to reinforce choosing the most appropriate test for a patient while being mindful of the radiation exposure and adhering to the principle of as low as reasonably achievable (ALARA).19,20 Due to their relatively short, physical half-lives, PET tracers may offer lower patient radiation doses than SPECT tracers. Our technique could expand the role of 13N-ammonia in many centers by overcoming the limitation of on-site cyclotron production. Treadmill 13N-ammonia PET can result in a lower effective dose than comparable Tc-99 m exercise protocols while maintaining the high diagnostic sensitivity of PET myocardial perfusion imaging.19,21 Radiation dose can potentially be further lowered using a stress-first protocol. Two low-risk patients in our cohort were studied according to a treadmill stress-first protocol. After review of their normal stress images, they did not undergo reinjection for resting images yielding an average dose of 16.5 mCi 13N-ammonia (approximately 1.6 mSv est. effective dose). According to estimates of effective doses of standard myocardial perfusion imaging protocols,19,20 this technique should result in similar or greater reductions in effective dose compared to low-dose, stress-only protocols using technetium 99 m SPECT and are comparable to the effective doses achieved in “ultra-low dose” protocols using Tc-99 m and CZT SPECT.20

Cost and Implementation

One of the limitations of the routine use of PET for MPI is its cost. PET imaging with 82Rb is associated with high cost due to its requirement of an on-site generator and regular, high volume use to make the generator cost-effective.22 Acquiring unit-dose 13N-ammonia for treadmill stress MPI may be more cost-effective but further studies and expansion of this dataset will be needed to determine scalability and economy of this technique. Future expansion of decentralized production of PET tracers with the use of small, in-laboratory cyclotrons7 may make tracers like 13N-ammonia more widely available for remote centers. Until that time, however, our technique offers a more immediate solution. Use of a remote cyclotron offers the ability for multiple hospitals in a metropolitan area to share a single cyclotron facility across a larger geographic area thus lowering the financial burden on a single institution. There are already ongoing clinical operations using remote cyclotrons for pharmacologic 13N-ammonia PET, and there are 101 commercial cyclotrons distributed around the US that could potentially be recruited for this purpose.23

Limitations of the Study

This is a single-center observational feasibility study with a relatively low number of patients. The study was not designed to test the relative diagnostic accuracy or extent/severity of perfusion defects with exercise compared to pharmacological PET MPI. The nature of the study precludes the ability to assess the cost-benefit ratio of obtaining unit-dose 13N-ammonia vs. generator-produced Rubidium-82. The protocol described herein may not be applicable to centers where there is no medical cyclotron within an appropriate driving distance.

New Knowledge Gain

Treadmill exercise 13N-ammonia PET is feasible in a large regional medical center with access to a nearby off-site cyclotron. In this limited study, image quality was consistently good and, in our opinion, superior to that seen in our large experience with 82Rb imaging. As exercise is the preferred method of stress testing in many clinical situations, the use of 13N-ammonia expands the potential for PET imaging to serve that population. It remains to be demonstrated that the approach can be cost-effective.