CHALLENGES TO THE SURVIVAL OF NUCLEAR CARDIOLOGY
Dennis L. Kirch
In considering the future prospects for nuclear cardiology we are confronted by a long list of issues such as diminishing reimbursement, increasing operating costs, unacceptable levels of radiation exposure, interrupted isotope supplies and tightened pre-certification requirements. While it doesn’t take long to define a half dozen external challenges to the field, an alternative view suggests that we should also take a critical look at some internal factors that have certainly contributed to the present state of affairs. The term “we” is used here in the collective sense here to include practitioners, academicians, technologists and the commercial collaborators who provide the products and services that the practice of nuclear cardiology depends upon. The specific issues raised here represent attitudes, assumptions and lack of rigor that have certainly exacerbated present circumstance.
Review of the facts underlying shortages in the supply of Tc-99m agents indicates that myocardial perfusion imaging (MPI) has significantly contributed to the crisis, largely by virtue of the sheer volume in the number of MPI studies performed each year. The manufacturer of Cardiolite proclaims in a long running journal advertisement (1) that “Since 1991 we’ve seen the population of New York City five times over”, so it shouldn’t come as a surprise that MPI is being scrutinized in terms of possible over utilization.
Tc-99m based MPI agents were heralded in the early 90's as cardiac tracers which increased productivity and were physically better suited for imaging with rotational-SPECT (R-SPECT) systems. This situation has now evolved to the point where pharmacy records indicate that the majority of nuclear imaging labs today routinely administer doses above manufacturer’s recommended levels. There is no published information indicating any demonstrable diagnostic benefit from increasing dosing levels. In response to the limited detection sensitivity of R-SPECT, Fast-SPECT hardware and software innovations (2) have demonstrated a doubling of throughput (or alternatively, a factor of 2 reduction in radiation exposure) and stationary multi-pinhole SPECT devices (2 and 3) have demonstrated an order of magnitude improvement in detection efficiency.
Compounding this problem is the industry-wide practice of pricing Tc-99m based MPI products by the “unit dose” rather than by the number of milliCuries administered along with the reimbursement allowance for passing through fixed charges to third party payers certainly dampens any financial incentive to use lower dosing. The promotion of “stress only” MPI seems an available means to reduce patient dosage but the inevitable reduction in diagnostic accuracy and inevitable adjustments in reimbursement will reduce enthusiasm for this maneuver in due time.
NEED TO REDUCE RADIATION EXPOSURE
Coincident with the recent radio nuclide supply shortages, concerns are now being raised by the FDA and others regarding acceptable levels of radiation exposure to patients, staff and the public. The clinical protocol most commonly applied for stress/rest MPI involves the use of the Tc-99m cardiac agent for both the stress and resting image acquisition. For many years an alternative dual-isotope R-SPECT protocol has been recommended by both the SNM and ASNC for same day stress/rest MPI. This approach involves the use of low-dose Tl-201 (<148 MBq) for the resting study followed by a higher dose (<740 MBq) of Tc-99m for stress. Note that this protocol recommends that Tl-201 be used as the resting agent. Again, this is a work-flow efficiency decision that is not supported by published reports on the diagnostic characteristics of the two tracers.
Strauss observed in a recent editorial (4) that there are demonstrable advantages to the use of Tl-201 over Tc-99m as the stress agent. The energetic properties of Tl-201 emissions are somewhat inferior to Tc-99m in terms of the resolution characteristics for imaging with conventional RSPECT gamma cameras using NaI(Tl) detectors. However, the higher extraction fraction of Tl-201 enhances flow differences between the stress and resting uptake of the tracer in flow compromised regions of the myocardium. Many reports comparing the use of these two agents for stress imaging (5) have concluded that the detection of stress induced flow differences is visually and quantifiably improved for Tl-201 in comparison to Tc-99m for MPI.
It is true that the longer physical half-life of Tl-201 corresponds to higher radiation biologic effectiveness (RBE) for Tl-201 in comparison with Tc-99m cardiac agents. Using fast SPECT technologies cited above (2) excellent quality MPI can be achieved using 74 MBq of Tl-201 as the stress agent and 185 MBq of Tc-99m as the resting agent. The resulting RBE for this low-dose, simultaneous dual-isotope regimen is 18.6 mSv/MBq which compares to 12.6 mSv/MBq for the single isotope R-SPECT protocol when adhering to the manufacturer’s recommended total (stress plus rest) dose of 1480 MBq of Tc-99m. It should also re considered that the RBE values for Tl-201 given above are significantly reduced when the physical half-life of Tl-201 (73.8 hr) is reduced to its biologic half-life (48 hr). With this adjustment the RBE for the single and dual isotope protocols become comparable.
Another factor contributing avoidable radiation exposure to patients, concerns the frequency of unnecessary cardiac catheterizations performed following falsely positive R-SPECT MPI. This fact is revealed from the diagnostic specificity associated with this procedure stated in the 2005 ASNC commissioned report on the cost effectiveness of R-SPECT MPI prepared by DePrez, et al (6). The overall sensitivity for R-SPECT MPI detection of CAD was stated to be 90% with specificity of 75%. Other sources corroborate these findings (7) leading to the conclusion that one in four normal patients undergoing MPI face possible unnecessary catheterization as the result of falsely positive MPI findings. At the present time the literature indicates that this situation has not improved for any of the commonly used R-SPECT protocols that acquire MPI in sequential, frame-mode (FM) fashion.
CURRENT STATUS OF MPI PERFORMED BY R-SPECT TECHNIQUE
To fully comprehend the challenges facing MPI using R-SPECT we need to review the premise that obstructive coronary disease manifests itself in terms of flow mediated differences in relative tracer uptake between the stress vs rest condition. These differences are demonstrated and quantified by comparing the “normalized” relative stress vs. rest tracer uptake delivered under these two conditions. In other words, MPI is a diagnostic modality in which the patient serves as their own standard and relative quantification requires that some specific segment of the myocardium be designated as “normal”.
This seemingly straightforward paradigm leads directly to the proposition that the presence of any systematic or procedural error in performing the individual stress/rest acquisitions will impair the comparability of these normalized flow differences. The existence of such errors has the potential to produce MPI interpretations which do not correlate with actual flow differences, thereby degrading the diagnostic accuracy of R-SPECT.
Results from the literature review cited previously (6) indicates that such errors are more likely to cause falsely positive MPI than falsely negative results (75% specificity compared with 90% sensitivity). To support reasonable cost/risk versus benefit expectations for MPI, both sensitivity and specificity need to be above 90% (7). A cursory review of current RSPECT technology points to some methodologic characteristics and protocol choices that certainly have the potential to cause systematic errors that could contribute to the disappointing diagnostic accuracy cited above.
ARTIFACTS THAT CAUSE FALSE POSITIVE R-SPECT MPI STUDIES
For example, R-SPECT acquires data over a prolonged sequential acquisition period during which the MPI data set generates a series of images taking from 10 to 30 minutes for acquisition. This characteristic indicates that any changes that alter the size and/or position of the heart in the field of view during the acquisition period can potentially create inconsistencies between the stress and rest MPI data set that will be manifested as quantifiable artifacts in the R-SPECT reconstructions. This circumstance was discussed in a recent editorial by Bacharach (8).
It is commonly thought that movement of the thorax, within which the heart sits, and “cardiac creep” due to tidal volume respiratory changes are the primary factors causing motion related R-SPECT artifacts. For instance, the phenomenon called “transient ischemic dilatation” (TID) is an accepted descriptor of an aberrant manifestation of this phenomenon. It is generally well accepted that, over an extended period of time following stress, the normal heart gradually returns to its resting volume and ejection fraction, which is typically occurring during the acquisition period of the RSPECT stress MPI. Therefore, it should be expected that the R-SPECT data set could often incorporate inconsistencies due to such volume changes into the MPI data set for the reasons cited by Bacharach (8).
PROTOCOL MODIFICATION FOR ELIMINATING SPECT ARTIFACTS
A simple but unwelcome adjustment to the R-SPECT acquisition protocol that will minimize the effects of these adaptive cardiac volume changes would be to have the patient rest in a supine position for an equilibration period prior to MPI acquisition. Unfortunately, this protocol would serve to allow any significant volume changes to resolve prior to commencement of acquisition. This approach was successfully used by Steele (9) to minimize (but not totally eliminate) such motion artifacts in order to objectively compare the results of R-SPECT and MP-SPECT. So it needs to be kept in mind that this equilibration maneuver minimizes our ability to observe TID which has been shown to be a valuable indicator for detecting the presence of multi-vessel disease (MVD).
Understanding the benefits of allowing a pre-acquisition supine equilibration period also sheds light on another misunderstanding about current MPI protocols. This concerns the suggested benefits of prone imaging. Previous work (10) has observed that prone stress MPI appeared to be “more clear” than the supine stress images. The source of this benefit has been attributed to the “reduction in subdiaphragmatic attenuation” afforded by the anterior shifting of the heart within the thorax provided by the prone position.
However, careful examination of the prone imaging methodology reveals that the validating protocol which was utilized for validation was flawed. In all of the published work on the benefits of prone imaging, the supine stress MPI was done first and the prone MPI was done second. Thus, there was a consistent beneficial bias favoring the prone image in terms of the “equilibration period” that was afforded by the duration of the supine MPI acquisition. Since the order in which the supine/prone images are acquired is never randomized, it leaves the possibility that the benefits of prone imaging have more to do with equilibration than attenuation. It should also be noted that, in the interest of maintaining stress/rest consistency and reducing systematic error, it would be advisable to “prone” the resting image as well as well as the stress image.
OTHER CONCEPTUAL MIS-STEPS IN R-SPECT INTERPRETATION
The term “reverse redistribution” has been in use for decades and was originally a descriptor applied to studies in which an apparent defect seen in the resting MPI diminished or vanished altogether from the stress images. No pathologic cardiac condition or circumstance encountered in clinical practice has ever been clearly and consistently identified as a satisfactory and consistent explanation for this phenomenon but a lucid discussion of the artifactual nature of the problem has been presented in an editorial by Borer (11).
This circumstance becomes more understandable when we recognize that both the stress and resting MPI are individually “normalized” prior to quantitative comparison. Traditionally, the normalization process is applied by treating the hottest myocardial segment as “normal” and adjusting all other segments in relation to this “normal” segment. The explanation for the appearance of “reverse redistribution” becomes clear once we consider that there is no metabolic, physiologic or anatomic compulsion that disallows that this “hottest” segment could not also be considered as a potentially abnormal location where flow differences could also occur. Typically, interpretation proceeds under the assumption that the hottest segment, with the greatest uptake, is the most normal region.
Once you include the possibility that flow differences can potentially be demonstrated in the hottest region of the myocardium, it becomes evident that the traditional “normalization process” will cause the intensity of the other, potentially “more normal” regions of the resting MPI, down to become relatively less intense than seen in the stress MPI. In this circumstance, correct quantification of the relative flow changes between the stress and resting MPI requires that the two sets of curves be normalized to a truly normal segment which we would expect to find in the valleys. This normal region would be at the point of greatest reverse redistribution and not at the maxima of the circumferential profile curves. Thus the correct process of quantitative stress/rest normalization depends very much on selection of a myocardial segment you can reliably define to be “normal”.
GROUP DATA BASE COMPARISON
An inevitable result of this conceptual gap in understanding the normalization process for quantitative stress/rest comparison of R-SPECT MPI has been the move to replace individualized relative stress/rest flow differences with extensive age/weight/sex/site specific group data base comparisons. In retrospect, it should be contemplated that a good deal of the motivation for developing group data base interpretation came about because of procedural, systematic and analytic artifacts. Often circumstances, prevent clinicians from trusting the reliability of comparing the patient to themselves (stress vs rest). It should be no surprise that the suggestion has now been raised that a resting study really isn’t even necessary because often this additional information adds “confusion” to the task of interpretation. The fact is that, without a true resting study for comparison purposes, it is difficult to detect those flow differences that occur in regions above the group derived lower limits of normal, thereby reducing the overall sensitivity of MPI for detection of low level CAD.
REASONABLE EXPECTATIONS FOR MPI IN THE FUTURE
The case being made thus far indicates that lack of rigor in the inclusion of known aspects cardiac physiology have combined with inherent deficiencies in R-SPECT performance characteristics and flawed MPI analytic protocols to create the circumstances which have inevitably degraded the diagnostic accuracy of R-SPECT MPI. The fact that these deficiencies have remained largely unacknowledged for so long fits in with the fact that many practitioners assume that R-SPECT in it’s current form can credibly serve the clinical needs of nuclear cardiology for years to come. It also explains why many of the recent developments being represented as improvements in SPECT MPI technology (2) only serve to increase the speed with which SPECT MPI is performed, leaving most of these deficiencies and the resultant diagnostic accuracy unchanged.
Gated MPI, introduced with high expectation several years ago, has proven somewhat beneficial in terms of visual correlation of wall motion/thickening abnormalities with the location of perfusion flow differences. At this time, lack of statistical content provides poor temporal resolution in terms of 8 frame/cycle gated MPI. The inability to directly and reproducibly measure wall thickening has relegated interpretation of gated MPI to be done visually. The desired goal in this endeavor should be to quantitatively correlate regions of compromised function in terms of reduced wall thickening and motion with regions of compromised blood flow.
Other shortcomings in R-SPECT efficacy include the difficulty of performing flow studies for quantification of cardiovascular flow reserve. The rapidity of tracer uptake and the duration of frame-mode (FM) step-andshoot acquisition is a longstanding dilemma (12). Also lacking is the flexibility needed to acquire gated studies during and immediately following pharmacologic stress using inotropic stress agents such as dobutamine. Such capability would add an echo-like dimension to gated MPI in terms of the ability to demonstrate regional wall motion defects as they develop during stress and then observe their resolution during recovery. In retrospect, it can be said that list-mode (LM) acquisition opens up many opportunities to resolve temporal, spatial, energetic and physiologic phenomenon that are completely masked by the commonly used FM stepand- shoot R-SPECT acquisition process.
Other capabilities that delineate the superiority of LM over FM for acquisition of MPI include the ability to correct for respirational motion (RM) which is a prerequisite to allow accurate analysis of wall thickening and wall motion. RM correction is also a prerequisite for high frequency, multiharmonic phase analysis of high resolution (>16 frames/beat) gated MPI. The present approach to phase analysis using maximum count analysis of center wall amplitude is indirect and imprecise and thus far has only demonstrated limited applicability for optimization of dual chamber pacing using statistical analysis of phase histograms (13).
IMPACT OF COMPETING MPI TECHNOLOGIES
Among the multitude of cardiac imaging technologies with capability for imaging the heart (which includes contrast ventriculography, coronary arteriography, Echo, cardiac MR, cardiac CT, PET and SPECT) , only PET and SPECT retain the unique characteristic of being molecularly based and therefore advantaged to reveal information about both the underlying myocardial metabolic condition along with quantification of ventricular performance. For this reason we are obligated to take a look at PET as an alternate approach to MPI which also provides guidance regarding a strategic pathway to substantial improvement in the quality of SPECT MPI.
In a competitive sense, PET technology for MPI has demonstrated advantages over the results achievable by R-SPECT. This includes the ability to perform dynamic flow studies for absolute quantification of coronary blood flow as well as high resolution gated studies of LV function. This is supported by the improved diagnostic accuracy comparing PET to RSPECT as reported by Bateman et al (7) which demonstrated 87% vs 79% for lesions > 70% and 87% vs 71% for lesions > 50%. Attempts to identify any technical, physical or physiologic basis for this result have alluded to aspects such as PET resolution, more robust attenuation correction, and reduced hepatobiliary uptake of positron emitting tracers as explanations. Note that these results were achieved in spite of the significant blurring associated with PET imaging of Rb-82 due its increased mean free path in comparison with F-18 or N-14.
Absent from the discussion of these possible explanations is any recognition of the most fundamental difference between R-SPECT and PET methodology. Specifically, the lack of detector motion by PET technique resulting in simultaneous acquisition of all of the image data comprising PET studies together. This characteristic of PET provides a consistent set of projection data in which motion or changes in the volume of the heart during acquisition affect all of the acquired image data equivalently. In this context motion has the effect of blurring the reconstructed images but does not cause artifactual perfusion defects.
Thus, in light of Bacharach’s observations (8) regarding how changes in the position and volume of the heart during R-SPECT acquisition have the potential to create artifacts, it becomes possible to make the more cogent observation that the reported improvement in the diagnostic accuracy of PET in comparison with R-SPECT MPI should more likely be attributed to the reduced presence of motion/volume induced artifacts incurred by PET due to it’s inherent simultaneous image acquisition characteristics. Using the delayed R-SPECT acquisition protocol previously suggested (9), this becomes a testable hypothesis.
ADVANTAGES OF MULTI-PINHOLE SPECT
The negative impact of R-SPECT artifacts is also supported in the review article commenting on the performance characteristics of multi-pinhole SPECT by Garcia and Faber (14). The fact remains that the only SPECT imaging modality that currently allows all of the single photon data to be acquired simultaneously without any mechanical motion is multi-pinhole SPECT (see pages 261-267 of reference 2). The multi-pinhole SPECT modality also fulfills the criterion for performing dynamic flow studies of cardiac perfusion which is mentioned by Gullberg (12).
At the present time there are two very different embodiments of multipinhole cardiac SPECT (2, 12). One of these is MP-SPECT+ which offers an upgrade path whereby an existing dual-detector SPECT camera with standard LFOV, NaI(Tl) detectors is modified to allow cardiac SPECT studies without any detector rotation or motion. This approach has been shown to support simultaneous MPI of Tc-99m injected at rest and Tl-201 injected at stress and has also been shown to be adaptable to a tripleheaded SPECT camera (9).
Applying a very different embodiment, the Discovery NM 530C multipinhole device utilizes the CZT based Alcyone TechnologyTM and has not been shown to support simultaneous Tc/Tl dual-isotope MPI capability. This difficulty is due to the space-charge build up phenomenon associated with CZT detectors which causes spectral down-shifting of primary events into the Compton region rendering scatter and cross talk correction between Tc- 99m and Tl-201 ambiguous. The simultaneous acquisition of stress/rest MPI data largely eliminates the systematic errors that plague R-SPECT because the stress and rest data are being acquired with the heart in the same position and size changes equivalently effect all views.
The intention here is to define the essential points necessary to provide a framework for evaluation of the current clinical practices and equipment performance characteristics that effect SPECT MPI. These comments also suggest future directions for technical development and clinical validation that will help strengthen and extend the role of clinical MPI by SPECT technique. From this review process it should also become clear that the information which is currently available in the literature is understandable, consistent and testable when correctly interpreted.
Also evident is the fact that candidate technologies to replace R-SPECT already exists. In order to secure a future role for SPECT MPI as a reliable diagnostic tool, the unique advantages of stationary, multi-pinhole SPECT MPI technique provide an example of the kind of methodologic advancement that will be needed to expand the role of MPI to provide additional diagnostic information, improve patient comfort and reduce the time, expense and radiation burden associated with MPI. While these are very commendable objectives, the primary objective remains to increase the diagnostic accuracy of SPECT MPI to acceptable levels of sensitivity and specificity (both >90%).
What is needed now is for innovative scientists, clinicians, technologists and manufacturers to move forward guided by a consistent and workable understanding of the principles underlying nuclear cardiology. The belief that R-SPECT MPI as it presently exists can adequately meet the future needs to detect and manage coronary artery disease is flawed and retention of this misguided notion will contribute significantly to further deterioration of the clinical role of MPI. As we enter an era of outcome guided medical management, simultaneous, dual-isotope MPI by multipinhole SPECT technique holds the key to accurate, reliable, cost effective MPI.
+ MP-SPECT is an upgrade product of Eagle Heart Imaging, Westminster, CO. The author is a part owner of Eagle Heart Imaging, LLC.
TM Alcyone Technology is a registered trade mark of General Electric Healthcare, Milwaukee, WI, www.gehealthcare.com/euen/fun_img/products/nuclear_medicine/products/ alcyone_technology/index.html
MP-SPECT by Eagle Heart Imaging, LLC