Evaluation of Diastolic Function by Radionuclide Techniques

Equilibrium Radionuclide Angiocardiography

Important variables for performing diastolic function analysis using ERNA are listed in Box 16.2. ^99m Tc pertechnetate is attached to the patient’s own red blood cells using one of three possible labeling methods, which vary in the time to perform labeling, expense, and labeling efficiency. Acquisition is conducted once the radionuclide attached to red blood cells has achieved equilibrium within the vascular space. Planar and SPECT methods of acquisition are available, but nearly all studies were and are being performed using planar techniques. Finally, the diastolic assessment can be conducted at rest and/or following stress with exercise (upright or supine bicycle) or a pharmacologic agent (dobutamine).

The patient is positioned in the supine or right lateral position for comfort, and a left anterior oblique (LAO) projection is obtained as the preferred one for diastolic assessment. This allows the best separation of the right and left ventricles for independent and accurate measurement of radioactivity within each chamber as a surrogate of volume. An example of a typical LAO view is shown in Fig. 16.4 . The labeled red blood cells circulate in the vascular space, and the patient’s ECG signal is used to set the timing for acquisition of each heart beat at individual time points, which may vary from 10 to 150 msec depending on heart rate and the preset parameters. Information for each time point for each beat (usually >400 beats are acquired) is summed so that the final TAC is an average rather than information from a single beat or a small number of beats, as is provided by other modalities. For this reason, this technique may be more representative of overall function than are other methods due to the large number of beats averaged.

An adequate temporal sampling of the radioactivity within the left ventricle is crucial to capture the minor or subtle alterations in diastolic filling. To achieve a temporal resolution less than 50 msec usually requires 16 to 32 frames per heart cycle. Therefore the use of 32 frames for each R-R interval is favored when analyzing diastolic function, especially at slow heart rates to appropriately sample all phases and phase transitions of diastole. Fig. 16.5 shows a 16-frame study, which can be contrasted to a study with 32 frame intervals shown in Fig. 16.6 . The higher temporal resolution obtained with 32 frame intervals results in the acquisition of even more detail during the diastolic filling period.

Unlike systolic function analysis, diastolic parameters are markedly affected by changes in heart rate and arrhythmias during acquisition and processing. To avoid filling inconsistencies due to heart rate variability and arrhythmias from such a large number of beats, the R-R interval is sampled prior to acquisition and the parameters set up to reject individual beats that vary by ±10% of the mean R-R. Since the beat following a rejected beat has a longer filling period and will add variability to the measurements due to differences in EDV and EF, it is also rejected.

There are three computer methods used to acquire ERNA studies: frame, list, and buffered beat acquisition. Frame mode is the simplest method. It samples the R-R interval prior to acquisition, determines the number of time frames for each beat, and puts data into the appropriate time frame in real time for each beat using the set parameters until a new ECG gating signal identifies the beginning of a new contraction. Once a beat has been added to the particular time bin it cannot be removed. If there is an early beat, a new ECG trigger resets the acquisition so that the time bins toward the end of the heart cycle will have fewer beats contributing counts. The TAC generated from this method of acquisition suffers from count drop-off in the terminal frame(s) due to heart rate variability. Although these TACs with terminal count drop-offs provide accurate EF measurements, as the TPFR and smallest ventricular volume tend to stay constant regardless of heart rate and contractility state, they cannot be analyzed for diastolic parameters using harmonic analysis. Despite the disadvantages of the frame mode, this method is universally used for assessment of systolic function because of its simplicity and the small storage size of the retained data.

List mode acquisition is considered the most accurate method. It retains the spatial distribution of the radiotracer for every millisecond of acquisition, along with the ECG trigger signal, so the data set can be reformatted after completion of acquisition into any timing interval and use the appropriate arrhythmia rejection during postprocessing. It provides maximal flexibility relative to heart rate variability but unfortunately at the expense of prolonged processing time and massive data storage. This technique was used exclusively for all studies performed by the group at the US National Institutes of Health (NIH).

A hybrid method, the buffered beat approach, is an attempt to provide realistic flexibility for heart rate variability while keeping data size to a manageable limit. It uses a temporary memory buffer to examine each beat and the ECG gating signal with regard to the set baseline parameters and makes an instant decision to keep or reject the beat. An additional feature of this method is a forward-backward curve generation technique to avoid discontinuities or count drop-off in the last several frames, which precludes harmonic analysis of the TAC.

Finally, the processed images are filtered to reduce statistical noise, the edges of the left ventricle are defined using manual or automated edge detection software, and background subtraction is performed. The radioactive counts within these boundaries is proportional to the total volume in the ventricle and are used to produce a TAC. A TAC from a frame mode of 16 time intervals acquired from a clinical study is shown in Fig. 16.5 . All the fine detail shown schematically in Fig. 16.2 is retained in this curve, which does not have any drop-off in counts in the terminal frames that may be seen in the presence of even minor sinus arrhythmias but is much more pronounced in the presence of atrial or ventricular arrhythmias.

First-Pass Radionuclide Angiography

First-pass techniques for assessment of diastolic function require administration of a compact intravenous (IV) bolus of radioactivity and use of a very high temporal resolution and high count rate camera system to follow the radioactivity as it traverses the right and left ventricles rather than after achieving equilibrium in the vascular space as with ERNA. Serial gated images at 25 to 50 msec of temporal resolution are acquired, and the resultant TACs allow systolic and diastolic function analysis of the right and left ventricles at rest or following exercise or pharmacologic stress. This technique often requires multicrystal camera systems to capture enough counts in the limited acquisition time of 8 to 10 beats during first pass. Such systems are no longer commercially produced. Newer cadmium-zinc-telluride (CZT) camera systems provide high sensitivity and spatial resolution and may potentially be used for FPRNA. The basic requirements for performing studies are shown in Box 16.3 and will be discussed in the following section.

Any ^99m Tc radiolabeled compound can be used for bolus administration. When assessment is to be made at rest and following stress on one day, an agent that is cleared by the kidneys ( ^99m Tc dimethyl sulfoxide [DMSO] or diethylenetriaminepentaacetic acid [DTPA]) can be given first, followed by an agent that is cleared by the liver ( ^99m Tc sulfur colloid) or a heart perfusion tracer ( ^99m Tc tetrofosmin or sestamibi), so that there is no interference during the second study from the initial dose. The total dose administered can be lower when a multicrystal camera is used, but with a single crystal camera that is not capable of high count rates, 25 to 30 millicuries (mCi) are required. The total dose must be given as a tight IV bolus over 2 or 3 seconds, which requires at least an 18-gauge IV in the antecubital fossa or the external jugular. If such venous access is not available, the study should not be attempted.

Anterior images are acquired with the patient supine or upright with the chest directly on the camera head. Exercise is best performed on a bicycle, which allows the chest to be relatively stationary and the images free of motion artifacts. Studies have been acquired during treadmill exercise, but this requires placement of external radioactive markers that can be used to correct for motion. Since only 8 to 10 beats can be used to analyze right ventricular (RV) or LV function without having bolus overlap in the chambers, heart rate variability or arrhythmias will further limit the number of beats that can be processed and will result in sampling bias or overlap of tracer activity in more than one chamber. These sampling limitations may result in poor study quality and limit the conclusions that can be reached.

ECG-Gated Perfusion Imaging Studies

Current ECG-gated myocardial perfusion techniques (SPECT and PET) offer the ability to adjunctively assess diastolic function using the concepts and methods from ERNA. For the assessment of diastolic function, ECG-gated perfusion studies use perfusion tracers (T1-201, ^99m Tc tetrofosmin or sestamibi, Rb-82) rather than Tc-tagged red blood cells utilized in ERNA, but similar methods of processing are utilized with these two radionuclide techniques. Unlike ERNA and FPRNA that identify temporal changes in radioactivity to assess ventricular function, ECG-gated perfusion studies track the ventricular endocardial borders using perfusion tracers, which allows both the evaluation of ventricular volumes and the assessment of systolic and diastolic function. To correctly perform diastolic function analysis from gated perfusion images, there are limitations to overcome. These include the low temporal resolution, failure to perform optimal arrhythmia rejection, and poor edge detection due to the low information density contained in the perfusion data. Despite these limitations there are several groups that have shown the feasibility of such an approach. Diastolic assessment using 32-frame gated SPECT correlated closely to evaluation obtained with ERNA studies. Although these systems are capable of generating values for diastolic function indices, these are not routinely used in clinical practice as the major interest is in perfusion and systolic function.

Unlike ERNA and FPRNA, in which the injected radioactivity remains in the intravascular space, ^99m Tc perfusion agents are cleared by the liver and empty into the gastrointestinal tract. Typically 20 mCi of ^99m Tc pertechnetate are injected and remain in the vascular space, whereas with the perfusion tracers 15 to 30 mCi are administered with only a small percentage taken up by the myocardium. This means that a smaller percentage of the injected dose is in the myocardium when imaging is performed and that the counts are low, mandating that the time intervals for each beat be relatively long to achieve adequate information density in each frame. Increasing the imaging time will improve total counts but decrease quality, as patients are likely to move. Most studies are acquired for 8 to 16 frames; however, acquisition at 8 frames might lack sufficient temporal resolution to adequately separate out and analyze the various components of diastole with low heart rates (125 msec/frame at a heart rate of 60 bpm). Conversely, an acquisition of 16 frames is optimal as it provides an improved temporal resolution of 63 msec at the cost of half the counts, yet adequate edge definition can still be performed.

The other problem that requires resolution is getting a TAC without terminal frame count dropout. Currently, arrhythmia rejection for ECG-gated perfusion studies uses ±50% of the mean R-R interval to avoid compromising the perfusion study, which would either drop beats and lower total counts or prolong the acquisition time to achieve adequate counts. Some of the newer camera/workstation systems allow multiple acquisition windows so that arrhythmia rejection can be performed on one set of data while a separate window can be used for the perfusion data without arrhythmia rejection. Additional time is not added to the acquisition. If arrhythmias are present or there is a poor ECG signal due to bad skin contact or muscle interference, the ECG-gated data set will still be compromised, and quality control needs to be performed.

The final hurdle is getting appropriate edge definition from the available algorithms to accurately track the endocardial borders of the ventricle. Existing software packages have been shown to handle the lower counts in 16-frame acquisition. Definition of the mitral valve plane at the base of the heart is also difficult, as the transition from the left atrium, which has little uptake, to the myocardium in the area of the valves and membranous septum is difficult to identify. Failure to clearly delineate this separation will result in error in the volume of the cavity and the resultant TAC.