AN ASNC 20TH ANNIVERSARY ARTICLE CME ARTICLE
Gated SPECT in assessment of regional and global left ventricular function: An update
Aiden Abidov, MD, PhD,a Guido Germano, PhD,b,c Rory Hachamovitch, MD, MSc,d Piotr Slomka, PhD,b,c and Daniel S. Berman, MDb,c
Gated myocardial perfusion SPECT (GSPECT) is a major clinical tool, widely used for per- forming myocardial perfusion imaging procedures. In this review, we have presented the fundamentals of GSPECT and the ways in which the functional measurements it provides have contributed to the emergence of myocardial perfusion SPECT in its important role as a major tool of modern cardiac imaging. GSPECT imaging has shown unique capability to provide accurate, reproducible and operator-independent quantitative data regarding myocardial perfusion, global and regional systolic and diastolic function, stress-induced regional wall- motion abnormalities, ancillary markers of severe and extensive disease, left ventricular geometry and mass, as well as the presence and extent of myocardial scar and viability. Adding functional data to perfusion provides an effective means of increasing both diagnostic accuracy and reader’s confidence in the interpretation of the results of perfusion scans. Assessment of global and regional LV function has improved the prognostic power of myocardial perfusion SPECT and has been shown in a large registry to add to the perfusion assessment in predicting benefit from revascularization.
Key Words: Gated SPECT Æ left ventricular function Æ diagnosis Æ prognosis Æ myocardial perfusion imaging
INTRODUCTION become a routine part of clinical protocols, expanding the clinical utility of myocardial perfusion SPECT. Over the last few decades, the assessment of ACC/AHA/ASNC guidelines for the clinical use of myocardial perfusion from stress and rest myocardial cardiac radionuclide imaging consider ECG-gated perfusion SPECT has become one of the most reliable SPECT as ‘‘…the current state of the art… The ability clinical tools in the management of patients with known to observe myocardial contraction in segments with or suspected coronary artery disease (CAD).1 ECG- apparent fixed perfusion defects permits the nuclear test gated SPECT, with the ability to measure LVEF and reader to discern attenuation artifacts from true perfu- ventricular volumes as well as to evaluate presence of sion abnormalities. The ability of gated SPECT to regional wall motion abnormalities (RWMA), has provide measurement of left ventricular (LV) EF (LVEF), segmental wall motion, and absolute LV
a volumes also adds to the prognostic information that From the Department of Medicine (Division of Cardiology), 2 The University of Arizona, Tucson, AZ; Department of Imaging can be derived from a SPECT study.’’ Most recent (Division of Nuclear Medicine),b and Department of Medicine (2012) multisociety guidelines for the diagnosis and (Division of Cardiology),c Cedars-Sinai Medical Center, Los- management of patients with stable ischemic heart c Angeles, CA; and the Heart and Vascular Institute, Cleveland disease support Class I and Class IIA recommendations Clinic, Cleveland, OH for diagnostic use of Gated SPECT in majority of Received for publication Sep 9, 2013; final revision accepted Sep 11, 2013. symptomatic patients with known or suspected CAD, Reprint requests: Aiden Abidov, MD, PhD, Department of Medicine and indeterminate or higher likelihood of either obstruc- (Division of Cardiology), Department of Medicine and Radiology, tive CAD (in case of no prior history of CAD) or The University of Arizona, Tucson, AZ; [email protected]. ischemia (in case of known prior CAD).3 J Nucl Cardiol 2013;20:1118–43. Gated SPECT (G-SPECT) is now performed in 1071-3581/$34.00 Copyright 2013 American Society of Nuclear Cardiology. nearly all myocardial perfusion SPECT studies in the 4 5-8 doi:10.1007/s12350-013-9792-1 United States and other countries. The strong appeal
1118 Journal of Nuclear Cardiology Abidov et al 1119 Volume 20, Number 6;1118–43 Gated SPECT in assessment of LV function
Gated SPECT as % of all SPECT from myocardial perfusion SPECT are not covered in 90 this review, and may be found in other recent reviews.11 80 70 60 HISTORICAL OVERVIEW OF TECHNICAL 50 AND CONCEPTUAL DEVELOPMENT 40 OF ASSESSMENT OF VENTRICULAR FUNCTION 30 USING GATED SPECT 20 10 Technical Milestones 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 The era of modern clinical nuclear cardiology began Year in the late 1960s with the introduction of the Anger scintillation camera, which was able to provide dynamic Figure 1. Growth in the percentage of all perfusion SPECT images of the cardiac distribution of radioactivity. Early studies acquired using the ECG-gated technique (courtesy of in the 1970s, the ability to accurately assess ventricular IMV Medical Information Division, Des Plaines, IL) (repro- duced with permission from reference9). function noninvasively, initially with end-systolic and end-diastolic equilibrium radionuclide ventriculography and the use of an ECG-gating device,12 and shortly thereafter with early medical imaging computers, led to the description and validation of the multiple gated acquisition (MUGA) scan, allowing nuclear cardiology to become a useful, routine clinical tool. At nearly the same time, nuclear cardiology found its first unique niche with initial clinical reports of the ability of nuclear cardiology to assess regional myocardial perfusion at rest and during stress with K-43 and Rb-81. While Rb- 81 was commercially available for a short time and used clinically for these measurements, the widespread clin- ical use of myocardial perfusion scintigraphy really began in 1976 when thallium-201 (Tl-201) became commercially available. The ‘‘thallium scan’’ rapidly Figure 2. Temporal trends in performing MPI as gated became the most widely used noninvasive imaging SPECT in Germany (reproduced with permission from reference5). method for detecting CAD in patients with an interme- diate likelihood of coronary artery disease and soon thereafter became widely employed for purposes of risk of GSPECT imaging is a direct consequence of the ease stratification of patients with known or suspected CAD. and modest expense with which perfusion assessment is During the late 1970s, exercise radionuclide ventricu- ‘‘upgraded’’ to include function assessment, and lography with both equilibrium and first-pass accounts for its becoming a routine part of myocardial approaches became another commonly performed stress perfusion SPECT (MPS) studies (Figure 1).9 The sig- imaging modality. By this time, nuclear cardiology had nificant temporal trend toward increase in proportion of emerged as being able to assess rest and stress myocar- MPS studies being performed as G-SPECT (Figure 2) dial perfusion and rest and stress regional and global has been seen globally over in the last decade.5 ventricular function, albeit with two separate studies. In This review is intended to describe the major several laboratories, such as at Cedars-Sinai Medical milestones in which ventricular function assessment has Center, these two tests were noted to often provide emerged and added to perfusion assessment using gated clinically useful complementary information.13,14 SPECT. Compared to our original 2006 JNC paper,10 we In the late 1970s, SPECT using a rotating Anger updated and added new references and figures describ- camera detector became widely available, increasing the ing most recent developments of the hardware and ability of myocardial perfusion scintigraphy to localize software utilized in the G-SPECT acquisition and and quantify regional myocardial perfusion defects.15 interpretation, and added important references advanc- However, assessment of ventricular function by nuclear ing our understanding of potential clinical implications cardiology still required the performance of a separate of the functional assessment using the G-SPECT meth- blood pool imaging study. G-SPECT was not widely odology. New findings relating to perfusion parameters used until 1990, when Tc-99m-sestamibi was approved 1120 Abidov et al Journal of Nuclear Cardiology Gated SPECT in assessment of LV function November/December 2013 for use in the United States. Its higher myocardial count demonstrated that utilization of these new dedicated rates (compared to Tl-201) improved image count cardiac cameras allow to reach lower radiation dose, and statistics, so that adequate images could be obtained significantly shorten acquisition time, leading to a with SPECT from the different parts of the cardiac substantial improvement in patient’s comfort without cycle, using ECG-gating.16 compromise in image quality.36-48 Strong correlations of The next important milestones in the technical the functional measurements with these new cameras development of the field were the widespread use of have been reported. multidetector cameras and dramatic increase in speed of Additional important technical milestone adding to computer systems, which made gated SPECT clinically the clinical importance of the gated SPECT is an ability feasible.9 By the mid 1990s, several centers had begun of (1) storing the large amounts of digital data, including to routinely perform GSPECT studies, allowing assess- raw datasets and reconstructed images, with capability ment of myocardial perfusion and ventricular function to easily retrieve this information as needed; and (2) from a single study. generating an automatic reports with possibility of A further highly important factor in the growth of storage of all the relevant clinical information in the GSPECT was development of the commercially avail- large data registries. Directly related to this ability, able software packages, allowing quick and automated several centers have gathered short and long-term quantification of parameters of both myocardial perfu- follow-up data49-51 and have also generated large sion and function. Totally automated ‘‘suites’’ of multicenter GSPECT registries. computer programs proved capable of providing simul- Analysis of these large databases has revealed taneous assessment of LV perfusion, global function important outcome data and crucial resource utilization (either systolic and diastolic), regional wall thickening trends, allowing assessment of cost effectiveness con- and motion, separate analysis of diastolic, systolic and siderations related to use of cardiac imaging in modern ungated datasets, as well as quantification of multiple clinical practice. ancillary parameters (LV mass, geometry, lung-heart ratio, transient ischemic ratio). Currently, there are Clinical Developments Related to Gated several commercially available gated SPECT computer SPECT software packages, including programs from Cedars- Sinai Medical Center (QGS—quantitative gated Clinical evolution of the method of GSPECT SPECT)17,18 University of Michigan (Ann Arbor, followed shortly after the development of technical MI),19-21 Emory University (Atlanta, GA),22,23 Stanford aspects of the field. Interest in GSPECT and its University,24,25 and Yale University (New Haven, CT)26 development was based both on its ability to assist with as well as some others. These software packages have technical artifact encountered during image interpreta- been validated in head-to-head comparison with clini- tion as well as to provide additional clinically useful cally proven imaging methods such as echocardiography, measurements of LVEF, volumes and wall motion MRI of MUGA (Table 1). In addition to ventricular simultaneously along with myocardial perfusion results function measurements, these software packages are able on a routine basis. The initial clinical use was to increase to quantify regional myocardial perfusion parameters as specificity in image interpretation. Myocardial perfusion well as the ventricular function measurements. For both defects that do not change between rest and stress are function and perfusion computer assessments, there has generally considered to represent areas of prior myo- been also extensive research demonstrating excellent cardial scarring. Due to non-uniform attenuation of repeatability of the results.27,28 radioactivity by tissue, reconstructed SPECT images in Most recently, vendors introduced additional novel patients without disease frequently show apparent non- noise reduction and/or resolution recovery protocols reversible perfusion defects, most commonly in the based on iterative reconstruction methodology29,30 lead- inferior wall in men and in the anterior wall in women. It ing to a significant image quality improvement and can be difficult from the perfusion scans alone to furthermore, allowing to obtain a high quality study with differentiate attenuation artifact from true fixed perfu- just a half-dose and/or half-time acquisition. sion defect. Gated SPECT improved this assessment by During last decade, we witnessed a revolutionary allowing the motion and thickening of the involved change in the hardware potential, related to develop- myocardial segments to be evaluated. Early in the ment, validation and successful introduction into a experience with gated SPECT, Taillefer et al reported clinical use of the ultrafast dedicated cardiac cam- that gated sestamibi SPECT studies in women were eras.31-36 From the very first publications of the small more specific than non-gated thallium-201 SPECT,52 single center trials validating protocols with novel solid presumably because the visual readers took into account state CZT detector cameras, multiple publications have the motion of segments with apparent fixed perfusion oue2,Nme 6;1118–43 Number 20, Volume Cardiology Nuclear of Journal
Table 1. A review of published quantitative algorithms for gated perfusion SPECT (reproduced with permission from reference9)
Institution Cedars-Sinai Emory U. U. of Michigan Stanford U. Yale U. Various others
Commercial QGSTM , AutoQUANTTM EGSTM , Cardiac 3D-MSPECTTM , MultiDimTM GSCQTM N/A name ToolboxTM 4D-MSPECTTM Operation Automatic Automatic Automatic Semi-automatic Automatic Semi-automatic Dimensionality 3-D 3-D 3-D 3-D 3-D 2-D Method Gaussian fit17 ,18 Partial volume22 ,23 Gradient19 ,21 Moment24 ,25 Maximal pixel, Partial partial volume73 ,74 ,237 volume26 Threshold 76 -79 Elastic surface80 Image inversion 81 Validation First pass,17 ,82 -85 First pass,23 MUGA,97 2-D First pass,132 First pass,85 First pass,81 of LVEF MUGA,85 -99 3-D MUGA,97 echo,122 MUGA,25 ,90 ,133 MUGA26 ,85 ,135 MUGA,77 ,79 ,80 ,97 ,136 MUGA,100 ,101 MRI,23 ,106 ,110 MRI,113 ,114, 129 ,130 2-D echo MRI,137 ,138 MRI,102- 114 2-D 2-D echo 121 contrast 133 ,134 2-D echo,121 echo,93 ,115, 117 -123 ventricul.20 ,21 Contrast 3-D echo,124 contrast ventricul.81 ,139 ,140 ventricul,100, 125 ,126 EBCT,127 thermodilution 128
Diastolic MUGA95 ,96 ,98 ,101 function LV of assessment in SPECT Gated parameters Volumes MUGA,90 ,91 ,93 ,97 ,99 First pass,23 MUGA,97 First pass,132 Excised MUGA,79 ,97 3-D MUGA,100 ,101 MUGA,97 MRI,19 ,113 ,114, 129 MUGA,90 hearts238 2-D echo,121, 239 MRI,2,102 -144 MRI,23 ,106 ,110 Excised hearts238 Contrast Contrast 3-D echo,3,93 ,115, 117 -123 3-D 2-D echo 121 ventricul.133 ventricul.140 echo,124 contrast ventricul,100, 125 ,126 EBCT,127 thermodilution,128 bdve l1121 al et Abidov excised hearts238 WM Visual18 MUGA240 MRI16 2-D echo241 WT Visual18 2-D echo144 1122 Abidov et al Journal of Nuclear Cardiology Gated SPECT in assessment of LV function November/December 2013 defects in the anterior wall. The concept employed is well as myocardial perfusion has been shown to further that if wall motion and/or thickening of a segment with enhance the ability of myocardial perfusion SPECT as an apparent fixed perfusion defect are normal, the defect a risk-stratification tool.58-60 is probably artifactual. Use of GSPECT for this purpose Subsequently, based in large part on publications is limited to the assessment of fixed perfusion defects, by Hachamovitch et al from the Cedars-Sinai database, because reversible defects due to stress-induced ische- the findings from gated SPECT have now allowed mia typically exhibit normal motion on the post-stress further transition into a paradigm based on prediction images (in the absence of stunning—see below). Even if of benefit from revascularization, rather than on risk the GSPECT finding does not actually change the alone.57,61-66 readers’ overall interpretation, it increases observer Taking in consideration functional variables derived confidence about the presence or absence of abnormal- from GSPECT, further enhances an ability of GSPECT ity. This phenomenon was first reported by Smanio to predict survival and adverse cardiac events; multiple et al53 and has subsequently been confirmed by others.54 publications to-date demonstrated incremental predic- Regarding increased reader confidence provided by tive value of wall motion and thickening scores, gated gating in the MPS interpretation, the recent ACC/ LV volumes, LV dyssynchrony indices, and other AHA/ASNC guidelines for the clinical use of cardiac ancillary markers on top of clinical and perfusion radionuclide imaging state ‘‘although this phenomenon variables as well as LVEF. Moreover, these functional is difficult to verify and quantify, it is reasonable to parameters remain prognostically important even in a expect that it would result in a reduction in the number subgroup of patients with abnormal LVEF.67 Recently, of ‘‘equivocal’’ studies reported.’’2 nomograms for CV risk prediction using combined Soon thereafter, the ability of GSPECT to define perfusion and function GSPECT data, were proposed by the presence of post-stress worsening of the LV the well-known group of experts in this field function, either segmental/regional or even global (if (Figure 3).68 severe and/or multiple RWMA were caused by the Below we will describe in details some of the stress) was described.55 This phenomenon representing technical and clinical developments of the field of exercise induced stunning became recognized as a GSPECT. marker of the presence of a severe stenosis, usually implying the presence of a critical (C90%) stenosis of GSPECT Acquisition Protocols the associated epicardial coronary artery, supplying the ’’stunned’’ segment.56 This finding formed the basis In a non-gated SPECT acquisition, the camera for the concept that stress ventricular function could detector(s) rotated around the long axis of the patient, add to the detection and risk assessment of patients acquiring one projection image at each of many, evenly with true abnormalities. spaced angular locations (steps) along the acquisition Arguably the most important clinical development orbit.69 With GSPECT acquisition, several (8 or 16, and and achievement of the entire field of nuclear cardi- recently even 32) projection images are acquired at each ology is that myocardial perfusion SPECT now plays a projection angle, with each image corresponding to a central role in mainstream clinical cardiology practice specific portion of the cardiac cycle termed ‘‘interval’’ as a means for risk stratification of the patient with or ‘‘frame’’ (Figure 4). The gating of a SPECT acqui- known or suspected coronary artery disease. Based on sition is easily implemented using the QRS complex of large databases, including clinical information of tens the ECG signal, since its principal peak (R wave) of thousands of patients who underwent nuclear imag- corresponds to end-diastole. The gating hardware inter- ing and were subsequently followed-up, robust faces with the acquisition computer that controls the evidence has emerged showing that MPS is effective gantry, and data corresponding to each frame are for this application. Initially, these results have sup- automatically sorted by the camera into the appropriate ported a change from a diagnosis-based approach to the image matrix. All projection images for a given interval decision about who should be considered for revascu- can be reconstructed/reoriented into a SPECT or tomo- larization to a risk-based paradigm in which nuclear graphic image volume using filtered backprojection or imaging plays a prominent role.57 With a risk-based iterative reconstruction techniques, and volumes relative approach, the focus is on identifying patients at risk for to the various GSPECT intervals can be displayed in major cardiac events, especially cardiac death, as well four-dimensional format (x, y, z, and time), allowing for as all-cause mortality, since to date only mortality the assessment of dynamic cardiac function. In addition, endpoints have been proven to be affected by coronary summing all individual intervals’ projections at each revascularization procedures. The ability of GSPECT, angle before reconstruction produces an ‘‘ungated’’ or with assessment of ventricular function parameters as ‘‘summed gated’’ SPECT image volume, from which Journal of Nuclear Cardiology Abidov et al 1123 Volume 20, Number 6;1118–43 Gated SPECT in assessment of LV function
Figure 3. Nomogram for estimating 2-year event-free survival (freedom from CAD death or nonfatal myocardial infarction). The nomogram combines prognostic information from the perfusion, function and type of stress test performed (reproduced with permission from reference68). perfusion can be assessed. Thus, GSPECT acquisition performed as early as 15 minutes after stress injection. yields both a standard SPECT dataset and a larger gated Adoption of this earlier post-stress imaging time has SPECT dataset. Of note, if counts were not included in most likely had the effect of increasing the frequency of the gated dataset because associated with heartbeats observing post-stress stunning, since it is known that this falling outside the ‘‘acceptance window’’, it is desirable process, by definition, resolves with time. While imag- for the standard (summed) dataset to contain them—this ing earlier than 15 minutes after stress is generally not can be achieved through use of an ‘‘extra frame’’ (a 9th recommended due to ‘‘upward creep of the heart,’’ it is frame in 8-frame, or a 17th frame in 16-frame GSPECT possible that with the application of effective motion acquisitions) where all rejected counts are accumu- correction algorithms,71 imaging could be started as lated.70 Initially, only post-stress gated acquisitions were early as 5 minutes after stress, further increasing the commonly employed. Over time, with increasing expe- frequency of observed myocardial stunning by rience and with increasing speed of the computer GSPECT.72 systems, it became the routine of most laboratories to perform gated SPECT both at rest and following stress, Computer Analysis of Gated Myocardial allowing for the study to be interpreted as showing post- Perfusion SPECT stress stunning through comparison of wall motion on the pre- and post-stress gated studies. Concomitant with the increased adoption of gated Initial protocols with sestamibi and tetrofosmin perfusion SPECT protocols, the past several years have were described as beginning one hour after stress tracer witnessed a substantial increase in the development and injection. Early after sestamibi became commercially use of algorithms for the quantification of global and available, Taillefer demonstrated that imaging could be regional ventricular function using GSPECT. In fact, it 1124 Abidov et al Journal of Nuclear Cardiology Gated SPECT in assessment of LV function November/December 2013
Figure 4. Schematic representation of ECG-gated perfusion SPECT acquisition and processing (reproduced with permission from reference9). is now estimated that essentially all gamma cameras of gating intervals are acquired95,96,101,145-147 (Figure 6). used for cardiac SPECT are directly connected to a Quantification of right ventricular function is generally computer or workstation running GSPECT software for not performed with gated myocardial perfusion SPECT function quantitation, at least in the United States. except in very special cases,148 but can be routinely Several software approaches for analysis of ventricular performed with gated blood pool SPECT, which is outside function from GSPECT are currently commercially the scope of this review. Regional parameters of function available. Table 1 presents a synopsis of published data quantified from gated perfusion SPECT images include on commercially available quantitative GSPECT algo- LV myocardial wall motion and thickening.149 rithms, including their principles of operation and the validation of parameters they quantitate.17,18,21-26,73-144 LVEF When combined with objective quantitative perfusion analysis approaches, the ability of these algorithms to Quantitative measurements of LVEF using gated provide automatic, operator-independent quantitative perfusion SPECT are usually volume-based rather than assessments of ventricular function and myocardial count-based methods. In particular, the time-volume perfusion at rest and after stress provides nuclear curve allows to identify the end-diastolic (EDV) and cardiology with one of its most distinct advantages over end-systolic (ESV) LV cavity volumes (Figure 7), from other noninvasive cardiac imaging modalities. which the ejection fraction is calculated as % LVEF ¼ ðÞEDV ESV =EDV 100 Majority of published validation studies of Standard Gated Myocardial Perfusion gated perfusion SPECT LVEF measurements by SPECT Study commercially available algorithms are presented in A wide variety of global and regional function Table 1, along with details about the variables can be measured from GSPECT (Figure 5). studies.17,21,23,25,26,82-97,103-108,113,115-134 Quantifiable global parameters of function from gated It is apparent that the agreement between GSPECT perfusion SPECT include LVEF, end-diastolic, and end- and gold standard measurements of LVEF is generally systolic LV cavity volumes, mass and transient ischemic very good to excellent. Indeed, it has been pointed out dilation (TID) of the LV based on gated or ungated that 2-dimensional gold standards may be intrinsically volumes. Diastolic function assessment was initially not less accurate than GSPECT algorithms operating in the thought to be possible with GSPECT, but published in last three-dimensional space, because of geometric assump- decade research shows it is feasible if a sufficient number tions required by the former.17 Journal of Nuclear Cardiology Abidov et al 1125 Volume 20, Number 6;1118–43 Gated SPECT in assessment of LV function
Figure 5. Diagnostic parameters derivable from gated myocardial perfusion SPECT study (reproduced with permission from reference9).
While most GSPECT measurements of LVEF in the literature have been derived from images acquired using 8-frame gating, 16-frame gating is becoming increasingly popular. Sixteen-frame gating requires additional data storage and processing time, and could result in images with unacceptably low-counts. How- ever, it can also provide more accurate estimates of LVEF, since there is more precise end-systolic imag- ing. Gated acquisitions with at least 16-frames are considered essential for diastolic function assessment. Quantitative measurements may also depend on the type of radionuclide used. As it was originally sug- gested with respect to non-gated dual isotope SPECT imaging (rest 201-Tl/post-stress 99mTc-sestamibi), res- Figure 6. Example of a patient’s volume and filling curves olution differences should be minimized by employing over time in 16-frame gated MPS. ED, end diastole, ES, end the same low-energy-high-resolution (LEHR) collima- systole, PFR, peak filling rate, BPM, beats per minute heart tor for the 201-Tl and 99mTc acquisitions.150 This rate, MFR/3, mean filling rate over the first third of diastole, approach is also advocated for GSPECT imaging. TTPF, time to peak filling; ms milliseconds. Numbers in brackets represent exact frame numbers from which the Nevertheless, it is to be expected that 201-Tl images parameters are derived. Arrow shows TTPF, defined by time will be more ‘‘blurred’’ compared to 99mTc-sestamibi from ES to the greatest filling rate in early diastole. Peak filling or 99mTc-tetrofosmin images, due both to the is normalized to EDV (reproduced with permission from 147 increased amount of Compton scatter associated with reference ). 201-Tl, and the use of a smoother pre-reconstruction filter. This would translate into a mild overestimation the QGS algorithm105,151-153 and shown in of 201-Tl LVEF and moderate underestimation of Figure 7 with respect to LVEF. Given the many 201-Tl EDV and ESV compared to 99mTc-based published validations of quantitative gated 201-Tl LVEF, EDV, and ESV, respectively, as reported for SPECT measurements against various gold 1126 Abidov et al Journal of Nuclear Cardiology Gated SPECT in assessment of LV function November/December 2013
Figure 7. The volumes bound by the endocardium and the valve plane at end-diastole (left column) and end-systole (middle column) are the highest and lowest point on the black time-volume curve (in this patient, frames # 1 and 7, respectively), from which the LVEF is calculated. The red curve is the derivative of the time-volume curve, from which parameters of diastolic function can be quantified (reproduced with permission from reference9). standards,87,89,90,92,105,106,117,120,124,133,153-155 it is unli- using these programs correlated well with the LVEF kely that major quantitative discrepancies exist values derived from equilibrium radionuclide ventric- between 201-Tl and 99mTc-based GSPECT, as long ulography.157 Modern postprocessing software is able as the studies are properly acquired and processed. This to reliably calculate LVEF even in patients with severe was also demonstrated in direct comparisons between perfusion abnormalities.158 separate151,153 and simultaneous dual isotope GSPECT acquisitions.83,156 Comparison of 4 commercially avail- Volumes able postprocessing software packages for the assessment of left ventricular ejection fraction (LVEF) It should be appreciated that one can usually more (Emory Cardiac Toolbox (ECTb), quantitative gated accurately measure ratios of LV cavity volumes, such as SPECT (QGS), four-dimensional single photon emis- the LVEF or the TID ratio,159 than the volumes sion computed tomography (4D-MSPECT) and themselves—for example, errors in the determination Myometrix) demonstrated strong correlation of the of end-diastolic and end-systolic volumes would be results among the programs; the LVEF values obtained expected to occur in the same general direction, and Journal of Nuclear Cardiology Abidov et al 1127 Volume 20, Number 6;1118–43 Gated SPECT in assessment of LV function therefore would at least partially cancel out when the is found in patients who have undergone bypass surgery; volumes are ratioed for LVEF calculation purposes.17 in these cases, ‘‘abnormal’’ wall motion with preserved Consequently, validation of quantitative LVEF mea- thickening of the interventricular septum is an expected surements does not necessarily imply validation of the normal variant. Similar discordance between wall end-diastolic and end-systolic volume measurements motion and wall thickening call also occur in left from which the LVEF is derived. Errors in the absolute bundle branch block, where preserved thickening with measurement of LV cavity volumes can be attributed, in abnormal motion of the interventricular septum is also a part, to the same factors that affect the measurement of common variant. At the edges of a large infarct, normal LVEF. Specifically, a) compared to 16-frame gating, the thickening with minimal or absent motion may be use of 8-frame gating will artificially increase end- observed in the peri-infarction zone, with reduced systolic volumes and have little effect on end-diastolic motion being due to the adjacent infarct. The presence volumes and b) end-diastolic and end-systolic volumes of thickening is considered to be indicative of viable will both be underestimated when quantitative analysis myocardium; conversely, ‘‘normal’’ wall motion of an is performed on unzoomed images of small ventricles, abnormally perfused segment that does not thicken especially with lower resolution radioisotopes. Above could be associated with passive inward motion of a considerations notwithstanding, a large body of pub- non-viable myocardial region (tethering), due to hyper- lished evidence suggests that quantitative measurements contractility of adjacent non-infarcted segments. of absolute LV cavity volumes from gated perfusion SPECT images agree well with established stan- Quantitative Wall Motion/Wall Thickening dards.23,90,91,93,97,99,100,102-113,115,117-119,121,123-125,128,129, 132,133,141-143,160 Assessment Quantitative wall motion and wall thickening assess- ment with interpretation software provides quantitative Visual Assessment of Regional Wall Motion methods for evaluation of the degree of wall motion and and Thickening wall thickening of each segment of the left ventricle; that Visual assessment of the LV RWMA has become might augment the visual analysis of ventricular function increasingly popular in modern nuclear cardiac imaging. from GSPECT data.27,149 Recently updated algorithms for The degree of wall motion is scored with a 6-point system the automatic quantitative measurement of segmental wall [0 = normal; 1 = mild (equivocal) hypokinesis, motion and myocardial thickening have been developed 2 = moderate hypokinesis, 3 = severe hypokinesis, and validated and demonstrate better performance com- 4 = akinesis, 5 = dyskinesis]. Wall motion analysis is pared to expert visual read (Figure 8).11,162,163 Moreover, performed by visualizing the endocardial edge of the left direct quantification of segmental rest-stress motion and ventricle, a process that is aided by the alternation between thickening changes may significantly improve accuracy of ‘‘contours on’’ and ‘‘contours off.’’ Many commercially GSPECT in diagnosing severe multivessel CAD164 available programs allow readers to utilize a three-dimen- (Figure 9). sional representation of the function data (‘‘views’’ in the QGS application), which provides three-dimensional con- COMBINED REST/POST-STRESS REGIONAL tours in three different interactive perspectives. FUNCTION ANALYSIS Visual assessment of wall thickening takes advan- tage of the direct relationship between the increase in the Modern computer software allows side-by-side apparent brightness of a wall during the cardiac cycle comparison of the rest and post-stress gated images to (partial volume effect74 and the actual increase in its identify the development of new wall motion abnormal- thickness. The degree of wall thickening is scored with a ity; this comparison becomes even more effective with 4-point system (0 = normal to 3 = absent thickening). new workstations’ dual-monitor displays. Wall motion During last decade we substantially improved our abnormalities that occur on post-stress images but are knowledge of the diagnostic value of the RWMA. It is not seen on resting images imply the presence of related not only to the technical development of nuclear ventricular stunning, and are highly specific for the cardiology, but also to developments in Cardiac MRI, presence of coronary artery disease.56105,106 Moreover, especially the understanding of the mechanisms of even if resting GSPECT studies are not available, ‘‘delayed hyperenhancement’’,161 and direct compari- presence of discrete post-stress RWMA can often be an son of the MRI and MPS images. In general, regional indicator of the presence of a severe coronary stenosis wall motion and wall thickening abnormalities accom- (C90% diameter narrowing). This finding might be pany each other. The most common cause of missed by perfusion defect assessment alone, particu- discordance between wall motion and wall thickening larly in patients with a greater degree of ischemia in a 1128 Abidov et al Journal of Nuclear Cardiology Gated SPECT in assessment of LV function November/December 2013
Figure 8. ROC curves for the detection of angiographically significant CAD disease by motion (M) (A) and Thickening (B) for individual observers (Obs1, Obs2), combined observers (Obs1 ? Obs2), automatic system (Auto), and previous system (Prev) (reproduced with permission from reference162). region other than that demonstrating the wall motion STUNNING abnormality.106 Johnson et al first reported in 1997 that of 61 patients with reversible ischemia imaged using a 2-day, treadmill stress and rest gated 99mTc-sestamibi SPECT REPRODUCIBILITY OF GLOBAL AND REGIONAL protocol, 22 (36%) had significantly lower post-stress QUANTITATIVE FUNCTION MEASUREMENTS LVEF compared to the rest LVEF.47 The threshold of The published results concerning the reproducibility ±5.2% (2 SD) for statistically significant differences had and repeatability of measurements of quantitative func- been determined from a separate group of 15 patients tion parameters from gated perfusion SPECT for undergoing serial rest gated SPECT on consecutive commercially available algorithms demonstrate very days. The authors attributed the reduction in LVEF in good to excellent agreement between independent mea- those 22 patients to postischemic myocardial stunning surements.17,18,25,55,82,91,93-95,97,99,111,119,132,152-154,165-174 persisting 30 minutes post-stress, noting that all 20 Even semi-automatic algorithms that require minor patients in yet another group without reversible ischemia operator intervention (slice selection, manual isolation demonstrated excellent agreement between the post- of the LV, manual identification of the LV cavity center, stress and rest quantitative LVEF measurements. etc.) generally enjoy equal or greater reproducibility The association between reversible ischemia and a compared to conventional nuclear or non-nuclear tech- decrease in the post-stress LVEF has been successively niques used for LV function assessment.175 Considering reported by a large number of investigators for exercise such a high reproducibility of the method, quantitative stress72,178-184 and even for pharmacological gated perfusion/function SPECT becomes an increas- stress.185-191 It has been suggested that subendocardial ingly important tool in the sequential evaluation of ischemia rather than stunning might be the main causing patients with progressing disease or undergoing medical factor for the apparent LVEF decrease, because quan- or surgical therapy.57,176177 titative algorithms might fail to adequately trace the Journal of Nuclear Cardiology Abidov et al 1129 Volume 20, Number 6;1118–43 Gated SPECT in assessment of LV function
Figure 9. Rest-stress myocardial perfusion 99mTc-sestamibi images of 53-year-old woman with history of hypertension and hyperlipidemia who underwent imaging to evaluate chest pain (A). Post-stress SPECT images clearly demonstrate reversible perfusion defect in left anterior and left circumflex artery. Motion and thickening change also identify abnormalities in right coronary artery territory (B). Cardiac catheterization confirmed presence of 3VD. STPD, stress total perfusion deficit (reproduced with permission from reference164). 1130 Abidov et al Journal of Nuclear Cardiology Gated SPECT in assessment of LV function November/December 2013 endocardium in regions with the greatest ischemia and severe and extensive CAD,159,212 with higher specificity thus underestimate LVEF192; however, studies designed than the LHR. Of note, multiple investigators have around sequential post-stress GSPECT acquisitions reported that TID ratio and LHR are not correlated (i.e., using 99mTc-based (non-redistributing) radiopharma- it is quite unlikely to find them both to be abnormal in ceuticals have shown that systolic dysfunction tends to any given patient),213,214 and suggest that these mea- resolve over time in spite of persisting stress perfusion surements may provide complementary information. defects, implying that true stunning is at least a partial The threshold for TID ratio abnormality depends on cause of the observed phenomenon.72,190,193-195 Inter- the choice of rest and post-stress radiopharmaceuticals, estingly, post-stress decreases in LVEF have been method of stress and possibly patient gender,215,216 but reported even in patients with normal perfusion,196 and appears to be relatively independent of the particular are associated with significant CAD56,181,190 and adverse quantitative algorithm used; published values range prognosis. from 1.14 for a same-day post-exercise/rest 99mTc- In addition to post-stress decreases in global LV sestamibi protocol216 to 1.22-1.23 for a rest 201-Tl/post- function, RWMAs present post-stress have been exercise 99mTc-sestamibi protocol,159,216217 with phar- described,56,185,197,198 and may be easier to detect macologic stress producing somewhat higher thresholds, compared to abnormalities in global post-stress func- as high as 1.36 with adenosine stress and dual isotope tion.56,183,185,199,200 Post-stress diastolic dysfunction has myocardial perfusion SPECT.212 It is also possible to also been found associated with systolic dysfunction in derive the TID ratio from end-diastolic or end-systolic patients with angina.201 LV cavity volumes,218 although the implications and While myocardial stunning is a well-recognized potential advantages of this approach have not been phenomenon both in conjunction with treadmill studied in depth. stress202,203 and adenosine vasodilator stress,189,204,205 there is less agreement on its duration. With respect to LV SHAPE quantitative GSPECT imaging, published reports range from less than 30 minutes195 to one hour or more with The LV can be reasonably approximated by an exercise stress198,206 or pharmacologic stress.185,207 The ellipsoid,17 and consequently it is easy to estimate its frequency with which post-stress decreases in LV shape using the major and minor axes of the ellipsoid function are encountered in a clinical laboratory will that best fits it. The closer the axes in size, the closer the likely depend to a large extent on the type of patients ellipsoid becomes to a sphere, a case consistent with LV imaged, with published data ranging from 5% to remodeling associated with congestive heart failure or 10%119,120,208,209 to as many as 44% of patients.196 In other pathologies. A potentially more accurate algorithm general, the finding is considered to be due to severe for shape assessment has also been proposed that is ischemia occurring during stress, and is usually associ- based on the regional search for the maximal distance ated with a critical ([90%) coronary stenosis.56 between endocardial surface points219 (Figure 10). Recently published data on post-stress ischemic stunning confirms previously published results and suggest importance of the stress-induced LVEF and volume change as an important prognostic marker.210 LV Shape Index TRANSIENT ISCHEMIC DILATION RATIO Normal sized Enlarged Enlarged non-remodelled heart non-remodelled heart remodelled heart TID of the left ventricle was first described for the A<