In this cohort study all patients that were referred to 13NH3 myocardial PET/CT were prospectively recorded in a database from December 2013 until November 2016. All patients gave written informed consent for the use of their data for scientific purposes. Besides our standard time-efficient imaging protocol and clinical management no additional measurements or actions affecting the patient were performed. Therefore, approval of the local ethical committee for the present study was not necessary since the study does not fall within the scope of the Dutch Medical Research Involving Human Subjects Act (section 1.b WMO, 26th February 1998).
Patients with a clinical history of CAD or signs of CAD on previous imaging studies were excluded. Patients with abnormal 13NH3 myocardial PET/CT results in terms of visual ischemia or infarction were also excluded, as were patients who showed coronary calcifications on the attenuation correction CT of the myocardial perfusion PET/CT. Patients with incomplete or visually suboptimal PET/CTs (e.g. due to motion artefacts) were also excluded. Patients displaying stress MBF below 1.8 ml/gm/min were excluded, since this somewhat arbitrary threshold value is considered below expected lower limits in previous literature (Gould et al. 2013; Hutchins et al. 1990) and could be caused by coronary microvascular disease.
The remaining patients were followed up after the 13NH3 PET/CT study until December 2016 by a thorough screening of the electronic medical records of each patient. In these records all emergency room presentations, hospital admissions, cardiac events, and cardiac and non-cardiac procedures in our hospital were recorded. Any events occurring in other hospitals are most likely recorded by our cardiologists during follow-up as outpatient. Patients were excluded if, according to these records, they suffered a major adverse cardiovascular event (MACE), defined as sudden cardiac death, non-fatal out of hospital cardiac arrest, acute coronary syndrome or atherosclerotic ischemic stroke.
Image acquisition and reconstruction parameters
All image data were acquired in list mode on a Siemens Biograph-16 TruePoint PET/CT (Siemens Healthcare, Knoxville, Tennessee, USA) with the TrueV option (enabled the axial field of view of 21.6 cm). This 3D system consists of a 16-slice CT and a PET scanner with four rings of lutetium oxyorthosilicate detectors.
After a topogram acquisition (110 kVp, 25 mAs) used for patient positioning, a CT transmission scan [130 kVp, 25 mAs (ref.), pitch 0.95] was acquired. Subsequently, patients were injected with 300 MBq of 13NH3 at rest. A 12-min rest imaging acquisition was started simultaneously with the start of the rest 13NH3 infusion (3 ml 13NH3, rate 0.4 ml/s followed with 17 ml NaCl and flushed with 20 ml NaCl at rate 2 ml/s). Pharmacologic stress was induced by an intravenous adenosine infusion (0.14 mg/kg/min for 6 min) immediately after acquisition of rest images. One minute after the start of the adenosine infusion, a 12-min stress imaging acquisition was started. Two minutes later, 400 MBq of 13NH3 was administered (3 ml, rate 0.4 ml/s). Immediately after completion of stress imaging, a second CT transmission scan was performed [130 kVp, 25 mAs (ref.), pitch 0.95].
Static, dynamic, and 16-bin ECG-gated images were generated from the list mode data. The acquired emission data were reconstructed using 3D attenuation-weighted ordered subsets expectation maximization (OSEM3D) reconstruction with a 168 × 168 matrix, zoom 2, a Gaussian filter with a full-width at half-maximum of 5 mm, two iterations, and 21 subsets for gated and dynamic images and TrueX (OSEM3D with PSF) reconstruction with a 256 × 256 matrix, zoom 2, a Gaussian filter of 4 mm, four iterations, and eight subsets for static images. CT-based attenuation, scatter, decay, and random corrections were applied to the reconstructed images.
Total scan duration was 25 min: 12 min for rest and 10.5 min for stress scans, with 2.5 min delay between these two parts. From each of the acquisitions dynamic, static and gated images were reconstructed. Dynamic rest images were reconstructed from the part of the dataset that was acquired directly after the initiation of the first 13NH3 injection, using 25 frames: 1 × 10, 12 × 5, 2 × 10, 7 × 30, 2 × 60, 1 × 180 s. MBF was subsequently computed by applying the Hutchins model to the dynamic images (Hutchins et al. 1990). For reconstruction of the gated and static rest images, the first 2.5 min of the rest acquisition data were skipped to allow for blood pool clearance and the subsequent 9.5 min of data were used. Gated images were used to assess left ventricular ejection fraction and regional wall motion abnormalities, while static images provide visualization of myocardial perfusion. The dynamic stress images were reconstructed from the part of the dataset that was acquired from 30 s before initiation of the second 13NH3 injection (which took place at exactly 15 min after the beginning of each scan) and onwards, using 26 frames for stress: 1 × 30, 1 × 10, 12 × 5, 2 × 10, 7 × 30, 2 × 60, 1 × 180 s. For reconstruction of static and gated stress images, the last 7.5 min of the acquisition data were used.
A cardiac specific motion correction algorithm from Siemens Molecular Imaging was used to detect the motion of the myocardium between the dynamic frames automatically. The motion was detected by a rigid image registration. The frames were re-aligned using an automatic motion correction method, propagating backwards from the final frame, comprising registrations between consecutive frames, until the frames no longer contained data that could be reliably registered due to change in image appearance as the sequence evolves.
Residual activity correction was applied for dynamic and static images obtained from the stress scan to exclude the interference of residual 13NH3 activity from the rest acquisition. Dynamic data sets were processed by subtraction of residual activity present in the first frame of the stress study (acquired directly before the stress 13NH3 injection) from the time activity curves using a residual correction method that was integrated into the Syngo MBF software package (Siemens Healthcare, Knoxville, Tennessee, USA). To assess the effect of this correction method, stress MBF values with and without residual activity correction will be provided. Static images were corrected by subtracting the residual activity measured at an interval of 2 min before the stress 13NH3 injection and corrected for decay and patient motion.
The quality of the registration between PET and CT was reviewed by experienced technologists and, in case of misalignment, corrected manually by means of 3D translations on the registration matrix before the final reconstruction process was started (Kan et al. 2016). MBFs were measured globally and for separate coronary artery territories and corresponding MFRs were calculated. Vital parameters were continuously monitored and rate-pressure products (RPP) calculated.
Statistical analysis was performed using SPSS software (version 20.0.0; IBM SPSS, Chicago, Illinois, USA). Categorical variables are presented as frequencies with percentages and continuous variables as mean ± standard deviation. Follow-up duration was described with median and interquartile range to better understand the distribution. Normal distribution was tested by the Shapiro-Wilk test. Data was subsequently divided by age and sex. Since most data was not normally distributed, significance between flow before and after residual activity correction was tested with the Wilcoxon Signed Ranks Test, significance of difference by sex with the Mann-Whitney U test, and significance of correlation by age with the 2-tailed Spearman’s rho test.