20 Nuclear medicine in heart failure

Nuclear medicine, sometimes known as molecular imaging, involves the characterization and measurement of biological processes in vivo using small amounts of radiolabelled tracers. It is the most sensitive imaging technique in routine use, providing images of nano- or even picomolar concentrations of the tracer. In patients with heart failure (HF), biological processes such as myocardial perfusion, metabolism (both fatty acid and glucose), injury (including necrosis and apoptosis), and innervation are relevant and can be imaged.

The ideal tracer is a biological molecule labelled with an isotope of one of its constituent elements (carbon, oxygen, nitrogen, etc.), since the tracer will have biological properties identical to those of the natural compound. Such isotopes are positron emitters and are imaged by positron emission tomography (PET), which relies on detecting the synchronous 511-keV photons emitted in opposite directions when the positron annihilates with an electron in the surrounding tissue.

More common tracers use foreign elements such as iodine or technetium bound to a pharmaceutical that provides useful biological properties. These radiopharmaceuticals usually emit single gamma photons that are imaged by a gamma camera, often using single photon emission computed tomography (SPECT) to produce tomograms or three-dimensional images.

PET has some inherent advantages over SPECT, such as higher resolution and more reliable attenuation correction that simplifies quantification of the biological process being imaged. PET is however more expensive and, with the exception of fluorine-18, the very short-lived radionuclides have to be generated on site, requiring the additional expense of a cyclotron.

Indium-111 antimyosin antibody imaging is very sensitive for the detection of myocarditis.¹ Myocyte necrosis leads to loss of sarcolemmal integrity and exposes insoluble molecules such as the heavy chains of myosin to the radiopharmaceutical, leading to a positive image of necrotic cells. Indium antimyosin has been used to detect cell necrosis in other settings such as acute myocardial infarction and transplant rejection² but, regrettably, this agent is no longer commercially available.

Gallium-67 imaging is used to diagnose several chronic inflammatory conditions and fever of unknown origin although it provides a nonspecific signal. It is useful in Lyme myocarditis³ and in the acute phase of Kawasaki’s disease.⁴ The incidence of histologically proven myocarditis is low (1.8%) in gallium-negative patients with dilated cardiomyopathy.⁵

After cell death, the cardiac myocyte does not regenerate and contractile loss is permanent.⁶ Noninvasive detection of myocardial injury is therefore important and imaging may be useful when ECG changes are absent or biomarkers cannot separate infarction from ischaemia.

Technetium-99m chelates such as pyrophosphates and glucarates have been used in the detection of myocardial infarction. Pyrophosphates accumulate in areas of necrosis through their binding to exposed mitochondrial calcium.⁷ A limitation, however, is that a positive signal is seen only after 24 h and up to several days and so the window of opportunity is relatively narrow.

Positively charged histones in the disintegrating nuclei of injured myocytes attract the negatively charged glucarate molecules and allow early imaging of infarction.⁸ Animal studies have shown that glucarate is taken up by necrotic and not ischaemic tissue and images can be obtained as early as 1 h in nonreperfused zones.⁹ Its rapid blood clearance and early uptake makes technetium-99m glucarate an attractive tool that is currently under investigation.¹⁰

In contrast to acute HF, there are several areas in the management of chronic HF where radionuclide imaging is routine, including diagnosis of the underlying cause in newly presenting HF, assessment of myocardial viability and hibernation in ischaemic heart disease, and investigation before resynchronization pacing and defibrillator implantation.

Myocardial perfusion scintigraphy (MPS) is an established and routine technique for the diagnosis of coronary artery disease.¹¹ There are fewer studies of its diagnostic accuracy in patients with left ventricular dysfunction but the sensitivity in this setting is extremely high.¹² Because nonischaemic myocardial scarring and inducible perfusion abnormalities also occur in primary and secondary muscle disorders, the specificity of the technique is not as high.^13,14 Features that suggest ischaemic heart disease are large areas of either transmural scarring or inducible ischaemia in a coronary distribution, impaired ventricular function in proportion to the loss of viable muscle, and predominant involvement of the left ventricle.

It is important to distinguish between myocardial viability and hibernation. Although these terms are sometimes used interchangeably, they are different entities. Viable myocytes are alive, irrespective of their ability to function, and viable myocardium contains viable myocytes. Viability is not a dichotomy and there is a continuum from fully viable myocardium with no scarring, through partial thickness scarring to transmural scarring.

Hibernation is an ischaemic syndrome whereby viable myocardium loses its ability to contract but is able to recover contractile function once the ischaemia is abolished.¹⁵ Hibernation was initially thought to be the result of reduced resting perfusion with contractile function being reduced in order to rematch oxygen supply and demand.¹⁶ It is now thought to be the result of repetitive episodes of ischaemia and stunning that mimic chronic reduction of function.¹⁷ It is likely that early hibernation is the same phenomenon as repetitive stunning but that changes in myocyte and myocardial structure later develop that may or may not be reversible. Hence, prolonged hibernating myocardium may lose its ability to recover and ultimately its viability through myocyte loss by apoptosis or necrosis.^18,19

In order to detect hibernating myocardium, all imaging techniques rely upon detecting a triad of signs: viability, function, and ischaemia (Fig. 20.1). The imaging definition of hibernation is therefore muscle that is viable but akinetic and where inducible ischaemia can be demonstrated. Without all three of these markers, even an area of viable but akinetic myocardium is unlikely to be hibernating and it could for instance simply be an area of partial thickness scarring without hibernation where the scarring is sufficient to abolish function.

Several radionuclide techniques can be used to detect myocardial viability and hibernation.^20,–22 They rely on the fact that all of the myocardial tracers are taken up only by viable cells, irrespective of the uptake mechanism. Hence, myocardial uptake reflects myocardial viability. In addition, tracers such as thallium-201, technetium-99m MIBI and tetrofosmin, nitrogen-13 ammonia, oxygen-15 water, and rubidium-82 have relatively high extraction and are fixed in the myocyte in proportion to delivery, hence they are dual tracers of viability and perfusion. When perfusion is less than 0.25mL min^–1 g^–1 or less than 30% of the blood flow in normal myocardium, recovery of function is unlikely after revascularization.²³ Preserved perfusion in areas of dysfunction indicates stunning.

Thallium-201, injected as thallous chloride, is a potassium analogue and its retention is an energy-dependent process reflecting myocyte viability and perfusion.²⁴ Conventional imaging for the detection of myocardial ischaemia involves injection during stress (dynamic exercise or pharmacological). Imaging within 30 min of this stress injection reflects a combination of myocardial viability and perfusion. After this time, the thallium equilibrates between the intra- and extracellular spaces and takes up a distribution that reflects viability alone, irrespective of perfusion. Defects in the stress images that improve after redistribution, therefore, indicate hypoperfusion at the time of injection, or areas with inducible ischaemia. Redistribution imaging is normally performed 4 h after the stress injection, although imaging up to 24 h allows even better detection of viable myocardium at the expense of lower-quality images as the thallium is excreted.

Using conventional stress-redistribution imaging, areas of dysfunctional muscle with an inducible perfusion abnormality are likely to be hibernating.²⁵ However, redistribution can be slow and the simple stress-redistribution technique can underestimate myocardial viability.²⁶ More sensitive techniques involve a further injection of thallium at rest after stress-redistribution imaging (the reinjection technique)²⁷ or a separate day resting injection of thallium with early and delayed imaging (rest-redistribution) (Fig. 20.2).²⁸ Resting injections are best given under nitrate cover (e.g. after sublingual glyceryl trinitrate) in order to abolish resting hypoperfusion and to give the best chance of thallium reaching any viable myocyte.

The technetium-based tracers MIBI and tetrofosmin have similar physiological properties to thallium although they have less avid extraction and are trapped within the myocyte and do not redistribute. Separate stress and rest injections are therefore required in order to detect inducible perfusion abnormalities. In contrast, technetium-99m has better imaging characteristics with a higher energy gamma emission (140 keV) and a shorter half-life (6 h). This leads to higher-resolution images, higher injected doses without excessive radiation exposure, and hence easier ECG gating for information on ventricular function.²⁹ The lack of redistribution means that viable but hypoperfused myocardium can be underestimated³⁰ and resting injections of MIBI and tetrofosmin should be injected under nitrate cover when assessing viability (Fig. 20.3).^31,–34 Clinically relevant viability has been defined as resting MIBI or tetrofosmin uptake of more than 50–60% of maximum but attenuation in regions such as the inferior wall means that a lower threshold may be relevant.³⁵

The myocardium is very flexible in choice of substrates for energy production, but β-oxidation of fatty acids and glycolysis are the two main mechanisms of ATP production.³⁶ Under aerobic conditions, fatty acid oxidation is the pathway of choice but under hypoxic conditions, glucose metabolism predominates because of its greater efficiency.³⁷

Fluorine-18 is a positron-emitting radionuclide with a half-life of 110 min and it can therefore be imaged by PET without the need for on-site production. As 2-deoxy-2-(¹⁸F)fluoro-d-glucose (FDG) it is widely used in oncology because of the affinity of rapidly dividing cells for the tracer. Uptake depends upon glucose transport, metabolism, and trapping within the cell and so FDG is a marker of glucose metabolism. Because normal myocardial metabolism uses fatty acids, if FDG is injected and imaged under fasting conditions there is no myocardial uptake but areas of glucose uptake indicate areas of current or recent ischaemia that have switched to glucose metabolism.³⁸ Alternatively, if FDG is injected after a glucose meal or during insulin and glucose infusion myocardial uptake is normal and defects indicate areas of scarring (Figs 20.4 and 20.5).

Straight-chain fatty acids can be labelled with radionuclides but their rapid metabolism complicates imaging. The β-methyl branched fatty acids are more suitable and the most widely used has been β-methyl-p-[¹²³I]-iodophenyl-pentadecanoic acid (BMIPP), although the tracer is currently only available commercially in Japan. When BMIPP is injected under fasting conditions it is extracted by the myocytes and converted into BMIPP-CoA, but it is not metabolized further. This results in high uptake and slow washout, making it very suitable for imaging.³⁹

Images are usually compared with thallium or technetium images. Concordant defects are scar, whereas discordant defects with a larger metabolic defect are likely to be hibernating.^40,41 Discordant BMIPP defects have enhanced FDG uptake as a result of the metabolic shift⁴² and have little fibrosis.⁴³ In a pooled study of 103 patients, 84% of segments with discordant BMIPP defects improved in function after revascularization but only 11% of those with concordant defects.⁴⁴

There are few large-scale studies of the accuracy of imaging techniques for predicting recovery of ventricular function after reva

scularization. Meta-analyses are confounded by different populations recruited, techniques used and imaging definitions of hibernation. For what it is worth, the weighted mean sensitivity and specificity for recovery of regional function after revascularization have been estimated as 92% and 63% for FDG PET, 87% and 54% for thallium MPS, 83% and 65% for technetium MPS, and 80% and 78% for dobutamine stress echocardiography. In general, all of the radionuclide techniques have similar accuracy and are said to be more sensitive but less specific than the dobutamine wall motion techniques.⁴⁵ In reality, though, the radionuclide techniques are more specific if the triad of signs referred to above is required before hibernation is diagnosed.

Fewer studies have looked at the improvement of global left ventricular function after revascularization but a similar trend is apparent in meta-analyses.⁴⁵ Fewer studies still have looked at more important measures such as symptoms and hard coronary events, but in nonrandomized observational studies patients with hibernation who are revascularized have better clinical outcome than those without hibernation, and the worst outcome is in patients with hibernation who are not revascularized. These observations can, however, only be hypothesis generating; randomized studies are needed to support the use of imaging in selecting patients with HF who will benefit from revascularization.

In the early stages of HF, increased sympathetic tone is helpful because it leads to increases in heart rate, contractility and venous return. As the syndrome progresses, this overactivity becomes unfavourable because of down-regulation and uncoupling of β-adrenoceptors, leading to progressive left ventricular dysfunction.⁴⁶ The sympathetic overactivity also increases the risk of arrhythmia mainly around focal areas of denervation.

Myocardial innervation can be imaged in a number of ways. [¹¹C]meta-hydroxyephedrine (HED) is an analogue of noradrenaline that concentrates in the presynaptic nerve terminal by the uptake-1 mechanism and provides high-quality PET images capable of quantification.⁴⁷  Meta-[¹²³I]iodobenzylguanidine (mIBG)is a guanethedine analogue with a similar mechanism of uptake that is suitable for both planar and SPECT imaging. Initial mIBG uptake corresponds with the density of sympathetic innervation and washout rate, thereafter, with the frequency of nerve terminal firing and hence sympathetic activity. Uptake is quantified from the ratio of myocardial and mediastinal uptake and washout is measured from changes between 15-min and 4-h images (Fig. 20.6).⁴⁸

Low myocardial uptake and rapid washout are associated with poor prognosis in both ischaemic HF and dilated cardiomyopathy^49,50 and myocardial uptake appears to be at least equal to left ventricular ejection fraction (LVEF) for predicting death and other major cardiac events.^51,52

Other applications of mIBG imaging may be the prediction of life-threatening arrhythmias and hence the need for an implanted defibrillator, since ventricular arrhythmias may arise from the border of an area of scarring with viable muscle that is denervated (Fig. 20.7).⁵³ In a study, of 50 patients with previous myocardial infarction and left ventricular dysfunction, significant abnormality on semi-quantitative mIBG SPECT, was 77% sensitive and 75% specific for predicting ventricular arrhythmias on provocation testing.⁵⁴ Patients with appropriate defibrillator discharges have lower myocardial mIBG uptake than those without.^55,56

The widest experience of the role of mIBG imaging for risk assessment in HF comes from the multicentre ADMIRE-HF study.^57,58 In this study, 964 patients with NYHA class II and III HF were studied with primary endpoints of HF progression, life-threatening arrhythmia, and cardiac death. There were 51 cardiac deaths in patients with heart to mediastinal ratio (HMR) of mIBG less than 1.6 compared with 2 in the high-uptake group, and 37% of patients with HMR less than 1.6 developed one of the endpoints compared with 15% with HMR greater than 1.6 (p 〈0.0001). The negative predictive value of high HMR for cardiac death within 2 years was 98.8%. HF death was more common in the lowest group but arrhythmias were more common with intermediate HMR (1.2–1.6), suggesting that focal as opposed to global denervation is more likely to cause arrhythmias and sudden cardiac death.

Until the prospect of myocardial repair by stem cells becomes a reality, prevention of myocardial injury and cell death should be an important objective in the management of HF.⁵⁹ Apoptosis is a form of enzyme-mediated cell death that is responsible for loss of myocytes in a number of conditions such as ischaemia and reperfusion, with up to 30% of myocytes becoming apoptotic after reperfusion injury.^60,61 The process can be inhibited, and this is cardioprotective.^62,63

Apoptosis is initiated by two pathways, intrinsic and extrinsic, with the final common step being activation of the caspase enzyme system.⁶⁴ Of the intracellular events that follow caspase activation, changes in phospholipid distribution in the cell membrane are central to apoptosis imaging. The maintenance of the lipid bilayer is energy dependent and one of the constituents, phosphatidyl serine (PS), is found in the inner layer. With caspase activation, the definition of the bilayer is lost and PS becomes externalized.⁶⁵ Annexin-V is an endogenous human protein with high affinity for PS that can be labelled with technetium-99m.⁶⁶

Technetium-99m annexin imaging has been successfully demonstrated in patients with recent myocardial infarction who showed annexin uptake 2 h after revascularization, corresponding with MIBI defects at 6 weeks.⁶⁷ In the HF setting, however, the main problem for imaging is the small number of cells undergoing apoptosis. In healthy hearts the prevalence of apoptosis is 1–10 myocytes per 10⁵ increasing to 80 to 250 per 10⁵ in advanced HF.⁶⁸ In a study of 9 patients with dilated cardiomyopathy, annexin uptake was seen in patients with worsening left ventricular dysfunction raising the prospect of using annexin imaging for prognostication.⁶⁹ In another study, 5 of 18 patients with recent heart transplantation had myocardial annexin uptake associated with at least moderate transplant reje

ction, suggesting the potential for annexin imaging to detect rejection and reduce the need for endomyocardial biopsy (Fig. 20.8).⁷⁰

Radionuclide ventriculography (RNVG) is a mature technique that provides an accurate and reproducible method of assessing ventricular function. It is sometimes referred to as MUGA (multigated acquisition) but the acronym is not sufficiently descriptive and RNVG is preferred. The intracardiac blood pools are labelled with technetium-99m, usually by in vivo labelling in which sequential injections of a stannous agent and technetium-99m pertechnetate are given. The stannous ion reduces the pertechnetate once inside erythrocytes and it is fixed intracellularly by complexing.

Images are acquired after the tracer is in equilibrium with the blood pool (the equilibrium technique) and/or during its first pass through the central circulation (the first pass technique). The former mainly provides information on left ventricular function but the latter can also be used to assess right ventricular function and left-to-right shunting. Traditionally, image acquisition is planar in the left anterior oblique projection that best separates the left and right ventricles but emission tomographic imaging is used increasingly. Acquisition is gated to the ECG to provide 16–32 frames over the cardiac cycle, with each frame containing counts from an average cycle over what is typically a 10-min acquisition for planar imaging. Images can be acquired at rest and during dynamic exercise but stress tomographic imaging is not feasible because of motion during the longer acquisition.

In the planar equilibrium technique, left ventricular count changes over the cardiac cycle are used to calculate ejection fraction and other parameters of both global and regional function, including diastolic function. Ejection fraction measurements require the subtraction of counts from structures in front of and behind the left ventricle and these are usually estimated from a region of interest lateral to the left ventricle. The estimation of background is a source of inter- and intraobserver variability but the technique is otherwise accurate and reproducible because it makes no assumptions about ventricular geometry. In the tomographic technique, ventricular volumes can be measured either geometrically or by summing counts within the ventricles.^71,72

Regional ventricular function is often assessed by Fourier analysis of count changes derived either from RNVG or from MPS, since myocardial count changes in MPS are related to myocardial thickening. Using either technique, Fourier analysis can also be applied to changes in geometry, for instance from the distance between the centre of the ventricle and the endocardium. Fourier analysis provides the parameters of phase and amplitude, which define the fundamental harmonic of contraction. Phase approximates to the time of end systole and can be expressed in milliseconds or more commonly as degrees or fraction of the cardiac cycle. Amplitude approximates to the magnitude of count changes through the cardiac cycle and hence to the extent of regional motion. Both parameters can be derived from whole ventricular counts, from regions or from individual image pixels. The histogram of phase values provides a simple method of displaying the mean and standard deviation of phase from any of these regions, and hence both inter- and intraventricular synchrony.

The combination of low LVEF and significant interventricular dyssynchrony measured by RNVG predicts improved systolic function 6 months after cardiac resynchronization therapy (CRT) (Fig. 20.9).⁷³ Similar analyses of MPS have shown a sensitivity and specificity of 70% and 74% respectively for predicting clinical improvement after CRT.^74,75