False-Positive Findings on Myocardial Perfusion SPECT

TO THE EDITOR: Recently, there have been 3 articles (1–3) published in The Journal of Nuclear Medicine attempting to define various artifactual sources that contribute to the unacceptably high rate of myocardial perfusion SPECT (MPS) studies with false-positive findings. Each article suggests a different protocol modification intended to reduce the incidence of false-positive MPS findings. The first paper (2) suggests a 15-min wait before initiation of poststress thallium SPECT. The second paper (3) concludes that 360° SPECT acquisition is superior to 180° acquisition, thereby doubling the acquisition time. The most recent paper (1) proposes that additional poststress prone images, commenced 20–40 min after reclining the patient for supine imaging, result in MPS interpretations that correlate more accurately with the clinical outcome.

We suggest an alternative and consistent explanation for the diagnostic improvements described in these studies. All 3 papers present protocol modifications that coincidentally result in an additional “equilibration” period during which the patient is supine. Thus, the diagnostic improvements demonstrated in all 3 of these papers could alternatively be concluded to be the result of the additional delay or prolongation that is inadvertently introduced by each of these protocols.

More thorough development of the methodologies described in all 3 papers could have redirected the authors to very different conclusions from those presented. We suggested alternative approaches in previous letters to the editor about the 2 earlier papers (2,3). In the case of the most recent paper (1), Hayes et al. could have conclusively demonstrated the validity of prone imaging if they had simply chosen to randomize the order in which the prone and supine imaging sequences were performed. Unfortunately, this work as presented indicates that none of their patients underwent prone imaging before supine imaging. Because the authors chose not to alternate the order of their poststress supine and prone acquisitions, we are left with the possibility that it was actually the 20-min delay before the onset of prone imaging, not the prone versus supine position of the patient, that gave rise to the benefits shown by their data. In a guest editorial that appears immediately following (1), Lee et al. (4) weakly support the work of Hayes et al. by indicating that they “might use additional prone imaging” until attenuation correction achieves “robust results.” Attenuation-induced artifacts are often described to be the nemesis of rotational SPECT, but neither attenuation correction nor prone imaging is widely used in practice because prolongation of the acquisition is not a welcome encumbrance in most busy clinical imaging laboratories. Furthermore, we submit that when the patient is reproducibly positioned and ^99mTc agents are used for both stress and rest images, and time is allowed after reclination for volumetric equilibration to be complete, there should be little if any artifactual stress/rest difference in MPS images due to attenuation. This leads us to wonder why additional resting prone images have not also been proposed as beneficial.

It also remains true that even the application of “robust” attenuation correction using sequential CT/SPECT transmission/emission myocardial perfusion tomography has not been shown to consistently eliminate the elusive “diaphragmatic attenuation” artifacts. All of this further supports our contention that another phenomenon is the dominant factor in generating false-positive MPS findings, namely, ventricular volume changes that occur during image acquisition when begun too soon after reclination of the patient. Positional changes (upright to supine) and poststress dynamic changes in the volume of the human left ventricle are well documented in the cardiac physiology literature.

In our previous work (5), we have graphically and statistically described the changes in left ventricular volume that occur during the 20 min following graded treadmill exercise and reclination of the patient for imaging. On the basis of these measurements, we propose that it is primarily these dynamic, positionally dependent ventricular volume changes that are the dominant factor generating artifacts when sequential, rotational SPECT images are reconstructed using standard, commercially available software. Until we fully comprehend the complexity of the myocardial perfusion imaging problem and design nonrotational SPECT systems (6) that accommodate its most demanding aspects, it will be difficult to advance the state of the art in nuclear cardiology to any higher level of clinical utility than is currently achieved by rotational SPECT systems.