Measuring myocardial blood flow using dynamic myocardial perfusion SPECT: artifacts and pitfalls

Abstract

Dynamic acquisition allows absolute quantification of myocardial perfusion and flow reserve, offering an alternative to overcome the potential limits of relative quantification, especially in patients with balanced multivessel coronary artery disease. SPECT myocardial perfusion is widely available, at lower cost than PET. Dynamic cardiac SPECT is now feasible and has the potential to be the next step of comprehensive perfusion imaging. In order to help nuclear cardiologists potentially interested in using dynamic perfusion SPECT, we sought to review the different steps of acquisition, processing, and reporting of dynamic SPECT studies in order to enlighten the potentially critical pitfalls and artifacts. Both patient-related and technical artifacts are discussed. Key parameters of the acquisition include pharmacological stress, radiopharmaceuticals, and injection device. When it comes to image processing, attention must be paid to image-derived input function, patient motion, and extra-cardiac activity. This review also mentions compartment models, cameras, and attenuation correction. Finally, published data enlighten some facets of dynamic cardiac SPECT while several issues remain. Harmonizing acquisition and quality control procedures will likely improve its performance and clinical strength.

Introduction

Adding dynamic acquisition to conventional static SPECT imaging using a dedicated cardiac CZT camera has the potential to be the next step of comprehensive perfusion imaging, by joining the best of both worlds. In this context, we sought to review the different steps of acquisition, processing, and reporting of dynamic SPECT studies in order to enlighten the potentially critical pitfalls and artifacts of this emerging technique. While Zavadosky et al provided a detailed overview of this topic,¹ our review focuses, in depth, on how to deal with the potential issues.

Two different dedicated cardiac CZT cameras are available for dynamic imaging: the D-SPECT (Spectrum Dynamics, Caesarea, Israel) and the Discovery NM 530c (GE healthcare, Milwaukee, WI). These cameras share a pixelated CZT detector with a cardiocentric acquisition geometry but vary in terms of collimation and reconstruction methods. The D-SPECT uses nine arrays of rotational detectors with parallel-hole collimation, whereas the Discovery NM 530c uses 19 immobile modules with multi-pinhole collimation. They both balance the resolution loss induced by the collimation with specific iterative reconstruction algorithms based on their collimators’ geometry. These different technologic options account for the main differences in performance between these two cameras: the Discovery NM 530c demonstrates a higher spatial resolution and the D-SPECT an increased sensitivity.² However, there are no data providing head-to-head comparison of these two cameras for myocardial perfusion quantification.

Wells et al previously detailed the technical aspects of dynamic acquisition.³ One of the main issue is the positioning of the patient prior to tracer injection and data acquisition. To ensure proper position of the patient, a positioning dose (40–70 MBq of ^99mTc-labeled tracer) is injected before the imaging session. Most studies in the literature were performed with patients in supine position, although some had patients in upright position.⁴ When available, the supine position is comfortable and appropriate to avoid patient motion during a relatively long imaging session. To prevent motion during the acquisition, it is very important to instruct the patient in detail about the different phases of the examination. Especially, the patient will be instructed to stand still and breathe normally during pharmaceutical injections. A repetition of this instruction may be given when starting the dynamic scan, 10–20 second before starting the radiopharmaceutical injection.

Dynamic acquisition requires pharmacological stress with regadenoson, dipyridamole, or adenosine. These stressors demonstrated similar results for stress MBF in normal volunteers using ⁸²Rb and ¹³N-NH₃ PET.5, 6, 7 Using dynamic SPECT, regadenoson and dipyridamole showed equivalent myocardial flow reserve (MFR) and stress myocardial blood flow (MBF) quantification after adjusting for clinical variables and cardiovascular risk.⁸ In a clinical workflow, regadenoson, despite a higher cost, has the advantage of (i) a single dosage (400 microg) in all patients, (ii) a bolus administration allowing a shorter total imaging time due to its pharmacological properties, (iii) a A2a receptor selectivity allowing its administration in patients with mild to moderate asthma. However, all three stress agents are reasonable choices for measuring myocardial blood flow using dynamic myocardial perfusion SPECT.

The first-pass extraction fraction is fair for ^99mTc-labeled tracers (54% for tetrofosmin, 68% for sestamibi).⁹ Thallium-201 does better with 85% but has its own drawbacks (mainly an increased radiation exposure and a higher cost). The difference between tracer uptake and coronary blood flow grows at high flows, which is a real impediment for quantification. The kinetic is corrected using the Renkin-Crone equation according to each radiopharmaceutical. As shown in Table 1, published data were acquired using sestamibi and tetrofosmin in similar proportions, and higher MFR cut-off for demonstrating myocardial ischemia was reported using sestamibi compared to tetrofosmin. Using a porcine model, Wells found a slightly better correlation with microspheres using thallium-201 compared to tetrofosmin and sestamibi.¹⁰ This difference may be related to the lower extraction fraction of tetrofosmin and sestamibi compared to thallium-201. See Figure 5 from Reference 10.

Using dedicated cardiac CZT cameras, the injected activity for dynamic SPECT is usually 3 MBq/kg at rest and 9 MBq/kg at stress. Acampa et al⁴ evaluated the diagnostic value of dynamic SPECT using lower injected doses (192 MBq at rest and 370 at stress) and reported a high sensitivity (86%) for the detection of coronary artery disease (CAD) verified by coronary angiography. However, it remains questionable whether reducing the injected dose will ensure an image quality suitable for a reliable quantification of dynamic studies, compared to PET.

Depending on clinical routine and institutional guidelines, tracer injection is performed either manually or using an automated injector. It is likely that the use of an automated injection would ensure a better repeatability of the bolus quality. Injection equipment must be carefully chosen, reliable injected activity being a key issue for quantification. In addition, attention must be paid to residual tracer activity. Swanson et al found a mean residual ^99mTc-sestamibi of 20% with a high variability (9%-31%).¹¹ This residual activity depends strongly on the type of syringe: Reynolds found 26% residual activity with standard design but 6% and 7% without silicone lubricant,¹² further encouraging the use of these latter syringes for dynamic acquisitions. The acquisition should start along with the bolus injection to allow a correct sampling of the bolus first pass.