Introduction

Radionuclide cardiac imaging, i.e., nuclear cardiology, is a well-established noninvasive method of evaluating patients with known or suspected heart disease. Its focus has been mainly on assessment of coronary artery disease (CAD) using myocardial perfusion imaging (MPI), enhanced by methods that evaluate left ventricular (LV) function and myocardial viability. The clinical utility of MPI has been well described and is broadly accepted.1-11

Nevertheless, heart disease encompasses more than CAD, and nuclear cardiology has much to offer beyond MPI. In particular, radionuclide imaging can assess molecular processes, helping to better understand the underlying cardiac pathophysiology, and thereby improving our ability to manage patients.12-14 A method under active investigation is imaging of the cardiac autonomic system that plays a major role in maintaining hemodynamic and electrophysiological stability at rest and in response to changing demands. In the setting of disease, autonomic control is often disrupted, with resultant image abnormalities that are both a reflection of disease severity and a prediction of further clinical deterioration.15 There is accumulating evidence that imaging with autonomic radiotracers can evaluate patients with a wide variety of cardiac conditions, including heart failure (HF), arrhythmias, and ischemic heart disease, providing highly effective risk stratification and therefore a potential guide for improving patient management.

Cardiac Autonomic Anatomy and Physiology

Cardiac autonomic control consists of both local innervation and circulating chemicals, and consists of the sympathetic and parasympathetic systems.16-18 The neurotransmitter of the sympathetic system is norepinephrine (NE), and that of the parasympathetic system is acetylcholine (ACh), working together to stimulate or inhibit the heart via adrenergic and muscarinic receptors. Sympathetic output is regulated by centers in the brain that integrate signals from other parts of the brain and receptors throughout the body (e.g., carotid sinus, aortic arch, origin of right subclavian artery, intracardiac). Efferent signals follow descending pathways in the spinal cord, synapsing with pre-ganglionic fibers and then paravertebral stellate ganglia, eventually innervating the right ventricle, and the anterior and lateral LV. In the heart, sympathetic nerves follow the coronary arteries in the subepicardium, before penetrating into the myocardium.

Parasympathetic fibers are relatively scarce. They begin in the medulla and follow the vagus nerves. They start epicardially in the heart, cross the atrioventricular (AV) groove, and penetrate the myocardium to be located in the subendocardium. Parasympathetic fibers predominantly innervate the atria, and are scarce in the ventricle (mostly the inferior wall), and also modulate sinoatrial (SA) and atrioventricular (AV) nodal function.

Published literature and clinical experience predominantly involve the sympathetic system, with parasympathetic imaging work reported mostly in animals. This review will therefore be limited to the former.

Development of Autonomic Radiotracers

Most autonomic radiotracers under investigation image pre-synaptic anatomy and function, illustrated in Figure 1 19,20 (although there is ongoing work on PET tracers that bind to post-synaptic α and β receptors). NE is produced in the pre-synaptic terminal, and is concentrated and stored in vesicles, often with ATP. In response to a stimulus, vesicular contents are released into the synaptic space and bind to postsynaptic receptors, resulting in cardiac stimulatory effects.21,22

Figure 1
figure 1

Schematic representation of the sympathetic neuron synapse. AC, Adenyl cyclase; AMP, adenosine monophosphate; cAMP, cyclic adenosine monophosphate; G, G proteins; NE, norepinephrine. Reprinted from Cardiology Clinics: Nuclear Cardiology—From Perfusion to Tissue Biology, Vol. 27, Travin19 Copyright 2009, with permission from Elsevier

Control of the sympathetic response occurs through a transporter protein-mediated, sodium-, and energy-dependent process, known as “uptake-1,’ (or norepinephrine transporter 1, i.e., NET-1) for storage and/or catabolic disposal, thereby terminating the stimulus. Some free NE is also taken up by non-neuronal postsynaptic cells, probably by sodium-independent passive diffusion (i.e., “uptake-2”).23,24

The development of autonomic radiotracers is well chronicled in a recent review by Raffel and Wieland. Cardiac tracer development was the unintended result of efforts by Dr William H. Beierwaltes, then Chief of Nuclear Medicine at the University of Michigan, to develop scintigraphic imaging for adrenal diseases, particularly neoplasms associated with the medulla.25 There researchers explored using iodine (I)-labeled analogues of the adrenergic blocking antiarrhythmic drug bretylium to image myocardial infarcts, finding that a para-iodo-bretylium analog also effectively imaged the canine adrenal medulla.26,27 The work continued with newly recruited radiochemist, Don Wieland, PhD, who investigated a related compound, guanethidine, a false neurotransmitter analog of NE. In 1979, it was determined that iodine in the meta-position yielded a tracer with comparable uptake to having iodine in the para-position but was metabolically more stable, and thus meta-iodobenzylguanidine, i.e., mIBG was selected as the optimal agent for adrenomedullary imaging.28 Cardiac mIBG uptake was also strong, with a significantly higher heart-to-blood uptake than the perfusion tracer then under investigation, 201Thallium (Tl). As per observations of Wieland, in contrast to thallium that depicted the “plumbing” of the heart, mIBG imaging allowed visualization of the “wiring” of the heart. The first publication of human heart imaging was in 1981 by Kline et al.,29 and shortly thereafter mIBG imaging was speculated as having the potential to assess the pathophysiology of HF, and identify patients with autonomic neuropathies who may be pre-disposed to arrhythmias and sudden cardiac death (SCD).30-32 Masyuki Nakajo MD, a visiting fellow from Japan, observed an inverse correlation between cardiac mIBG uptake, and plasma and urinary catecholamine levels,33 and after returning to Japan found that intraneuronal uptake of the tracer into storage vesicles was a key process.34 Shortly thereafter, efforts at Michigan shifted toward positron emission tomography (PET) compounds, resulting in clinical mIBG studies thereafter being performed mostly in Europe and Japan.

While early in development, mIBG was labeled with 131I, high energy emissions (365 keV), including β particles, and the 8-day half-life led to use of 123I that emits predominantly 159-keV gamma photons, with a half-life of 13.2 h, thus well-tolerated and easily imaged with single-photon emission computed tomography (SPECT). Unlike NE, after uptake in the pre-synaptic terminal via the NET-1 pathway, 123I-mIBG is not catabolized, and thereby localizes to a high cytoplasmic concentration.29,35

Imaging Procedure and Interpretation

Cardiac imaging with 123I-mIBG has been used clinically in Europe and Japan for years, but at the time of this writing is only US FDA approved to image pheochromocytomas and neuroblastomas (under the brand name AdreView™). There is no established standard for tracer administration and imaging. Flotats et al,36 with the European Association of Nuclear Medicine and the European Council of Nuclear Cardiology, recently proposed such a standard.

Tracer Administration

123I-mIBG is performed at rest with only minimal preparation. Standards for a recent multicenter study were to keep the patient NPO.37 Based on several studies it is accepted that standard HF medications such as β-blockers, angiotensin-converting enzyme inhibitors (ACE-I), and/or angiotensin receptor blockers (ARBs) need not be held.38-40 However, it is recommended to temporarily discontinue medications and substances known to interfere directly with the mechanism of NE uptake, such as opioids and cocaine, tricyclic antidepressants, sympathicomimetics (e.g., ephedrine, pseudoephedrine, phenylephrine, isoproterenol), some antihypertensive and cardiovascular agents (e.g., labetolol, reserpine, bretylium, calcium channel blockers), antipsychotics (e.g., phenothiazines), and foods containing vanillin and catecholamine-like compounds (e.g., chocolate and blue cheese).41,42

There are differing views regarding pre-test administration of thyroid blocking agents. Historically such blockade had been undertaken to shield the thyroid from exposure to unbound radionuclide iodine impurities, but with modern production methods the amount of these is minimal, and many feel that pre-treatment is unnecessary. For now, pre-treatment should be based on local and institutional regulations.37

Earlier studies administered a dose of 3-5 millicuries (mCi) (111-185 megabecquerels (MBq)) over 1 minute. As it is often difficult to obtain satisfactory SPECT images using these doses, especially in patients with severe cardiac dysfunction, investigators have recently been using up to 10 mCi (370 MBq).22,37

It has been recommended that patients lie quietly in a supine position for at least 5 minutes before administration. As initial images are acquired a few minutes later, the tracer is best administered slowly over 1-2 minutes while the patient is under the camera or in close proximity.

Adverse reactions to 123I-mIBG are uncommon. Among side effects reported when administered too quickly are palpitations, shortness of breath, heat sensations, transient hypertension, and abdominal cramps. A rare anaphylactic reaction is also possible. A 10 mCi dose results in radiation exposure of ~5 mSv, with highest exposure to the bladder, liver, spleen, gall bladder, heart, and adrenals; the absorbed dose may be higher in patients with severe renal impairment.36

Imaging Technique

123I-mIBG is currently imaged using a standard Anger gamma camera, with a symmetrically centered energy window of 20% around the main 159-keV isotope photopeak. Most clinical and published work use a low energy high-resolution (LEHR) collimator, although slight differences among collimators in different countries (US vs Europe vs Japan) yield slightly different normal quantitative values. However, because of septal penetration by higher energy (>400 keV) 123I photons, mostly a 529-keV emission, some have recommended using a medium energy collimator that has been shown to provide superior quantitative accuracy.36,43,44 To compensate for the resulting corruption of image quantitation related to LEHR collimator penetration by higher energy photons, Chen et al45,46 have developed a mathematical technique (iterative reconstruction with deconvolution of septal penetration) that appears to improve quantitative accuracy of cardiac 123I-mIBG uptake in reference to a phantom standard. Clinical applications have yet to be determined.

Much data about 123I-mIBG imaging is based on analysis of planar images, mostly a standard anterior view. Planar images are typically acquired 15-20 minutes after tracer injection (early image) with the patient supine, and again 4 hours later (late image), for 10 minutes each. SPECT images can also be acquired using standard perfusion imaging methods.36 Eventually other acquisition, processing, and display techniques may be developed given issues such as often extremely poor cardiac uptake in patients with advanced HF, frequent uptake in adjacent lung and liver that overlap myocardial walls, and the need in many cases for absolute as opposed to relative quantitative myocardial tracer uptake.

As interpretation of planar images requires analysis of the upper mediastinum, there is concern that smaller field of view cardiac cameras, or performing SPECT imaging alone, would not be suitable for cardiac 123I-mIBG imaging. However, a recent study by Chen et al.47 reported development of a technique to derive satisfactory quantitative parameters for SPECT imaging limited to the cardiac field

Image Analysis and Interpretation

Analysis of cardiac 123I-mIBG images consists of quantitative analysis of global uptake, i.e., the heart-to-mediastinal ratio (H/M); the difference in tracer uptake/retention in early and late images, i.e., the washout rate (WR); and regional uptake on SPECT images, often in relation to uptake in separately obtained standard perfusion images. Methods for all analyses remain under investigation.

At least three methods have been described to obtain an H/M ratio. In one, squares or rectangular regions of interest (ROIs) are drawn in the center of the heart and upper mediastinum, with a count per pixel ratio calculated.48 In another, an ROI is drawn around the epicardial border and the valve plane, including the LV cavity.49 Finally, some use an ROI encompassing the myocardium alone, tracing the epicardial and endocardial borders, excluding the valve plane and cavity.50 Interestingly, all methods appear to give similar result. Figure 2 illustrates a method recommended in the European guidelines,36 with counts/pixel in the myocardial ROI divided by count/pixel in a mediastinal box located above the lung apices, below the thyroid gland. Techniques are being explored to standardize the H/M ratio, with Okuda et al51 describing an algorithm that automatically determines the mediastinal ROI based on tracer uptake in the heart, lung, liver, and thyroid. Normal values for H/M range from 1.9 to 2.8, a mean of 2.2 ± 0.3, with a ratio <1.6 (2 SD below mean) investigated as indicative of possible increased patient risk.35,37 Figure 3 shows examples of normal and abnormal global cardiac 123I-mIBG uptake.

Figure 2
figure 2

Method of H/M ratio determination. Counts per pixel in the myocardial heart (H) region of interest are divided by those in the mediastinum (M). Reprinted from Flotats12,36 with kind permission from Springer Science and Business Media

Figure 3
figure 3

Examples of planar cardiac 123I-mIBG images. The example on the left shows normal cardiac 123I-mIBG uptake with a H/M ratio of 2.24 and a normal tracer washout (WO) rate from initial to delayed images (not shown) of 10.64%. The example on the right shows abnormal cardiac 123I-mIBG uptake with a H/M ratio of 1.29 in images and an abnormal tracer washout of 23.35%. Reprinted from Ji and Travin20 with kind permission from Springer Science and Business Media

123I-mIBG washout, i.e., the difference in cardiac activity between early and late planar images (compensated for radioactive decay), may reflect turnover of catecholamines attributable to sympathetic drive, and measures the ability of myocardium to retain tracer. The level of circulating catecholamines may also affect washout.52 A normal value has been reported to be 10% ± 9%.53,54 Increased sympathetic activity, reflecting worsened HF, is associated with diminished myocardial 123I-mIBG retention on delayed images and thus a higher myocardial WR.55 Although various methods of washout determination are reported, recent European guidelines indicate the following36:

$$ \begin{aligned} {\text{WR}}_{\text{BKGcorrected}} = \frac{{\{ H_{\text{e}} - M_{\text{e}} \} - \{ (H_{\text{l}} - M_{\text{l}} ) \times 1.21\} }}{{(H_{\text{e}} - M_{\text{e}} )}} \times 100, \end{aligned} $$

with 1.21 the correction for 123I decay at 3 hours and 45 minutes, e the early images, l the late images, BKG the background, H the heart counts per pixel, M the mediastinal counts per pixel, and WR the washout rate.

Interpretation of tomographic 123I-mIBG images is less well established, in part because of frequent poor quality, as well as variations in normal individuals. The rationale for SPECT is that the presence of regional autonomic tracer defects, particularly if tracer uptake is relatively preserved on a separately obtained standard rest perfusion image, i.e., an autonomic/perfusion mismatch, may indicate potential for electrical heterogeneity and denervation supersensitivity, pre-disposing to potentially lethal arrhythmias.56,57

While there is no officially established method for scoring SPECT 123I-mIBG images, analysis can be performed similar to the conventional 17-segment method used for MPI, with generation of a summed score.37,58 However, a key difference for 123I-mIBG images is that when there is globally decreased uptake, homogeneous tracer uptake cannot be scored as normal as, unlike the custom for perfusion images, one cannot assume a “normal” region.59 A software program that incorporates the issue of globally decreased uptake has been developed and is being tested for the Emory Cardiac Toolbox application (personal communication Ernest V. Garcia, PhD, and Russell D. Folks, CNMT). Another problem relates to frequent overlying extracardiac (lung and liver) activity that can obscure parts of the myocardium.

Regional 123I-mIBG uptake can be heterogeneous in healthy individuals. Somsen et al54 observed lower activity in the inferior than the lateral wall possibly from anatomic variation in sympathetic nerve activity. Heterogeneities may be more pronounced in men and healthy older subjects.60 Estorch et al61 found 123I-mIBG uptake to be lower in the inferior wall in athletes with sinus bradycardia, perhaps from increased vagal tone. Positron emission tomographic (PET) tracers such as 11C-hydroxyephedrine (HED) and 11C-epinephrine also show normal heterogeneity, but less so, suggesting that there are both physiologic and technical issues involved.62

Clinical Applications

Heart Failure

The most investigated clinical use of cardiac 123I-mIBG imaging is in patients with HF, a condition of high morbidity and mortality affecting >6 million American over age 20.63 As HF largely involves disruption of the neurohormonal state, including activation of the renin-angiotensin-aldosterone system (RAAS) and compensatory activation of the sympathetic nervous system (SNS), cardiac neuronal innervation is thought to play a key pathophysiologic role.64 An increased sympathetic response in HF patients with reduced cardiac output leads to deleterious neurohormonal and myocardial structural changes that worsen the condition and increase the likelihood of a poor outcome. Initially a compensatory attempt to maintain cardiac output leads to increased NE release, promoting the NE transporter 1 (NET-1) process. Eventually the NET-1 system is overwhelmed, with a reduction in NET-1 carrier density, leading to increased spillover of NE into plasma, likely accounting for the increased washout seen on 123I-mIBG imaging in patients with HF. With progression of cardiac dysfunction there is diminished pre-synaptic function from loss of neurons and down-regulation of NET-1, likely accounting for decreased cardiac uptake (lower H/M) in advanced disease.52

Following the initial report by Kline et al of human cardiac 123I-mIBG imaging, Schofer et al65 were the first to describe a potential role for 123I-mIBG imaging in HF, finding decreased cardiac uptake in 28 patients with idiopathic dilated cardiomyopathy that correlated inversely with LVEF, but surprisingly did not relate to circulating catecholamine. Prognostic utility was first reported in a 1992 landmark study by Merlet et al of 90 patients with advanced HF (NY Heart Association (NYHA) Class II-III symptoms and LVEF <45%), finding that H/M was superior to and independent of cardiac size on chest x-ray, echocardiographic end-diastolic diameter, and LVEF in predicting survival.66 An H/M < 1.2 was associated with 6- and 12-month survivals of 60% and 40%, respectively, while all patients with H/M ≥ 1.2 survived despite severe HF.

Subsequent work by Nakata et al67 of 400 patients showed the utility of H/M as a continuous variable, with progressively worsening survival as the H/M decreased, with H/M again a more powerful predictor of outcome than other conventional HF variables such as NYHA, age, prior myocardial infarction (MI), and LVEF. Following accumulation of similar findings in several single-center small trials, Agostini et al performed a 290 patient, combined data reanalysis study from 6 European sites, showing that the only significant predictors of major cardiac events over 2 years were LVEF and H/M. As in Figure 4, particularly striking is the ability of H/M to risk stratify patients with LVEF ≤ 35% in a continuous fashion, with event rates ranging from <5% for those with HMR ≥ 2.18 to over 50% for those with HMR ≤ 1.45.68

Figure 4
figure 4

Major cardiac event rates (MCE) over 2 years in relation to left ventricular ejection fraction (LVEF) and 123I-mIBG H/M. Cardiac events include cardiac death, transplant, and potentially lethal arrhythmias based on implantable cardioverter defibrillator discharge. Reprinted from Agostini et al68 with kind permission from Springer Science and Business Media

Cardiac 123I-mIBG WR has been investigated by Ogita et al, showing that patients with washout ≥27% had a 35% 4-year cardiac death rate compared with no deaths for a normal WR, and a threefold increase in HF hospital admissions in the high WR group.55 Another study from this group reported that increased WR predicted SCD.69

Upon meta-analysis of literature from 18 prior studies, a total of 1,755 patients, Verberne et al reported that abnormal WR had a pooled hazard ratio (HR) of 1.72 (P = .006) for cardiac death, and a HR of 1.08 (P < .001) for cardiac events (cardiac death, MI, transplant, HF hospitalization); in the three best studies reported for late H/M, there was a HR of 1.82 (P = .015) for cardiac death and 1.98 (P < .001) for cardiac events.70

Efforts culminated in the AdreView Myocardial Imaging for Risk Evaluation in Heart Failure (ADMIRE-HF) trial, a prospective, multicenter, international study of 961 patients with NYHA Class II-III and LVEF ≤ 35%.71 At 17-month follow-up, an H/M < 1.6 more than doubled—from 15% to 37%—the incidence of worsening NYHA class, life-threatening arrhythmias (sustained ventricular tachycardia >30 seconds, resuscitated cardiac arrest, and appropriate implantable cardioverter defibrillator (ICD) discharge), and cardiac death (CD), with a composite hazard ratio of 0.40 (P < .001) for a higher H/M. Multivariate analysis showed that H/M was a predictor of cardiac and all-cause deaths independent of other clinical and image variables, including age, NYHA functional class, LVEF, and brain natriuretic peptide (BNP). In particular, there were only 2 CDs for 201 patients (about 20% of total) who had H/M ≥ 1.6, including for ejection fraction (EF) <20%, thus a low negative predictive value of <1%, shown in Figure 5.72 Although only one study, findings from ADMIRE-HF indicate that 123I-mIBG imaging in otherwise high-risk HF patients can identify a significantly large subgroup who are in fact at low risk, at least over an approximately 2-year follow-up.

Figure 5
figure 5

Relationship of left ventricular ejection fraction (EF) and H/M to 2-year cardiac mortality in the ADMIRE-HF study. Reprinted from Chirumamilla15 with permission from Elsevier

In a subanalysis of ADMIRE-HF patients, Ketchum et al73 found that H/M added significantly to the prognostic power of the Seattle Heart Failure Model (SHFM),74 an algorithm of routinely collected demographic, imaging, laboratory, and therapeutic parameters that determine the likely 1-5 year mortality. Adding H/M to the SHFM-D algorithm (modified by data from the SCD-HEFT -Sudden Cardiac Death in Heart Failure Trial)75 yielded a net reclassification improvement of 22.7%, with 14.9% of subjects who died reclassified as higher risk, and 7.9% of patients who survived reclassified as lower risk, shown in Figure 6.73

Figure 6
figure 6

2-Year-mortality risk reclassification enhancement when 123I-mIBG H/M is added to the Seattle Heart Failure model (SHFM-D). The net re-classification improvement from image findings was 22.7%. Reprinted from Ketchum et al73

Assessing Response to Therapy

Given concerns about overuse of medical testing, it is important that the risk stratification ability of a modality such as 123I-mIBG imaging lead to improved patient outcome. Recent American College of Cardiology Foundation/American Heart Association HF guidelines recommend comprehensive pharmacologic regimens.76 As mortality for CHF patients remains high,77 when pharmacologic therapy is insufficient advanced mechanical device therapies such as biventricular pacemakers for cardiac resynchronization therapy (CRT), left ventricular assist devices (LVAD), and implantable cardiac defibrillators (ICD) should be considered, as well as cardiac transplantation. To help better decide the need for advanced therapies, ways of assessing pharmacologic efficacy that might include a surrogate endpoint, such as improvement in an imaging study, should be useful. Much work has shown potential for 123I-mIBG imaging to be effective in this regard. For example, numerous studies have shown that cardiac 123I-mIBG images improve after therapy with β-blockers.48,50,78-83 Gerson et al50 showed that the H/M improved significantly after the use of carvedilol, especially in patients with an H/M ratio <1.40. Toyama et al84 showed favorable changes of symptoms, functional class, cardiac function, and H/M in those treated with metoprolol. Kasama et al85 reported on the therapeutic effect of carvedilol on 123I-mIBG parameters and LV remodeling in patients with dilated cardiomyopathy.

Although improvement in autonomic function parameters in response to β-blockers is understandable, other medications, such as angiotensin-converting enzyme inhibitors (ACE-I), angiotensin receptor blockers (ARBs), and spironolactone that affect the renin-angiotensin system, also improve cardiac 123I-mIBG uptake.86-91 Amiodarone, an antiarrhythmic medication that would not be expected to directly influence cardiac sympathetic function, has also been shown to improve 123I-mIBG parameters in patients with advanced HF.92

Many ask if 123I-mIBG imaging might help direct medical therapy. Such investigations in terms of β-blockers have shown that imaging does not provide sufficient separation between those who do or do not benefit,83,93 and given the high benefit/risk and relatively low cost of medical therapies, an 123I-mIBG study is unlikely to preclude their use.94 123I-mIBG imaging could instead be used to determine whether or not a particular therapy is working, perhaps increase doses more aggressively or determine if device therapies or transplantation are needed.95,96 Matsui et al97 studied patients with severe cardiomyopathy, and found that after 6 months of optimal medical therapy, a worsening H/M had, with BNP, the highest predictive value for CD, suggesting that such patients may have benefited from earlier device therapy or transplant. At the same time, a recent study by Drakos et al98 of patients with an LVAD found that clinical improvement paralleled improvement in tracer uptake, suggesting that 123I-mIBG could guide which patients need transplant, or instead who might be able to have the device discontinued. Along the same lines, it is reported that a decreased H/M predicted poor response to CRT.99

Cardiac Arrhythmias

A major cause of mortality in HF patients is SCD, most often from a ventricular arrhythmia.100 SCD is particularly tragic in HF patients who otherwise have a reasonable quality of life. The currently accepted approach to potential SCD is an ICD, sometimes as secondary prevention after an aborted event, but increasingly more as primary prevention without specific evidence of risk. Recommended ICD use for primary prevention derives mostly from four large randomized studies: Multicenter Automatic Defibrillator Implantation Trial-II (MADIT-2),101 Defibrillator in Acute Myocardial Infarction Trial (DINAMIT),102 Defibrillators in Nonischemic Cardiomyopathy Treatment Evaluation (DEFINITE),103 and SCD-HeFT.74 HF guidelines assign a Class IA recommendation for implantation of an ICD as primary prevention in patients with NYHA Class II-III symptoms and LVEF ≤ 35%.76

LVEF has become a major variable for deciding who should receive an ICD, but this approach is flawed. One issue is that while the aforementioned trials indicate good relative survivals and significant P-values for ICD benefit, the absolute decreases in mortality are fairly small, from about 5.6% to 7.2%, with 11 to 14 patients needing an ICD to save 1 life.74,101,104 This degree of benefit must be balanced against substantial risks and costs of an ICD.105-108 The randomized trials have limitations, in particular MADIT-II and SCD-HEFT having broad enrollment criteria with limited stratification of study populations.109 Differences among EF entry criteria were large, and most enrolled patients had EFs well below the threshold ultimately used in guidelines. Buxton et al110 found that multiple factors other than LVEF provide more accurate prediction of SCD and mortality. Over half of the patients who die suddenly have an LVEF > 30%,111-113 and thus guidelines do not recommend an ICD for the majority of patients who have SCD. In part, because of perceived guideline limitations, many clinicians are not following them.114,115 Lack of clarity about a patient’s true LVEF, often based on visual estimates “subject to bias and reader error,” often differing depending on the imaging method chosen, create more uncertainty.109,116 Many feel that a better method of deciding on an ICD as primary prevention is needed.117

Autonomic imaging depicts cardiac pathophysiology closer to the underlying mechanisms of arrhythmias,100,118 and there is much evidence that 123I-mIBG imaging can effectively indicate which patients are likely to benefit from an ICD.119 Arora et al.,120 in a small study of 17 patients with advanced HF and an ICD, found that an H/M < 1.54 was associated with increased incidence of ICD discharges, and that on tomographic imaging patients who had ICD discharges had more extensive 123I-mIBG defects and more extensive autonomic/perfusion mismatches, shown in Figure 7. An example of a SPECT images in a patient with severe/extensive 123I-mIBG defect(s) and autonomic/perfusion mismatch is seen in Figure 8. Subsequently, Nagahara et al121 prospectively followed 54 patients with an ICD, finding that H/M correlated significantly and independently with appropriate discharges and SCD. Nishisato et al reported that a combination of H/M and the summed perfusion defect score on 99mTc-tetrosfomin images separated patients with ICD shocks from those without, with image variables independent and superior to age, sex, SAECG, BNP, medications, inducible arrhythmias, and LVEF in predicting shocks or cardiac death.122 Kasama et al123 showed a correlation of abnormally high 123I-mIBG washout with increased SCD. Tamaki et al. compared ECG parameters—HRV, QT dispersion, and SAECG—with 123I-mIBG findings in 106 patients with LVEF < 40%, and those with SCD had a lower H/M and higher WR, with ECG variables showing no independent relationship to outcome.124 In ADMIRE-HF, combined arrhythmic events were more common in subjects with H/M < 1.60 (10.4%) than in those with H/M ≥ 1.6 (3.5%, P < 0.01).71 In a subanalysis of 578 patients without an ICD, Senior et al125 reported only one fatal arrhythmic event in patients with H/M ≥ 1.60. In terms of tomographic imaging, Bax et al59 reported that in patients with prior MI, the extent/severity (summed score) of 123I-mIBG defects correlated with electrophysiological VT inducibility, and Boogers et al. found that in HF patients with a mean LVEF of 28% who received an ICD, a summed score >26 independently predicted more frequent ICD discharges and cardiac death (13-fold higher risk).126 Interestingly neither of these latter two studies found a correlation of autonomic/perfusion mismatch with pre-disposition to arrhythmias.

Figure 7
figure 7

Planar and SPECT 123I-mIBG results in relation to the occurrence of implantable cardioverter defibrillator (ICD) discharges in 17 patients with ICDs and 2 control patients without heart disease. Compared with patients who did not have an ICD discharge (ICD− patients with a discharge (ICD+) had a lower mean HMR, a higher mean neuronal tracer defect score, and a higher mean neuronal tracer uptake/perfusion tracer mismatch score. Reprinted from Cardiology Clinics: Nuclear Cardiology—From Perfusion to Tissue Biology, Vol. 27, Travin19 with permission from Elsevier

Figure 8
figure 8

123I-mIBG and 99mTc-sestamibi SPECT images of a patient who had received numerous appropriate ICD shocks. There are neuronal/perfusion mismatching defects involving the inferior, inferolateral, and apical walls; there is a matched defect in the anterior wall. HLA, Horizontal long axis; ICD, implantable cardioverter defibrillator; MIBG, metaiodobenzylguanidine (123I-mIBG); MIBI, 99mTc-sestamibi; SA, short axis. Reprinted from Ji and Travin20 with kind permission from Springer Science and Business Media

Thus, there are consistent findings that 123I-mIBG images can predict ICD discharges and SCD independent of conventional variables. In particular, a satisfactory H/M has an extremely high negative predictive value. Nevertheless, it is understood that larger prospective studies are needed before there can be wide acceptance and inclusion of 123I-mIBG imaging in guidelines.127 At the same time, cardiac neuronal imaging could potentially identify a subgroup of patients thought of as lower risk (e.g., LVEF > 35%), but who are instead at significant risk of SCD and may need an ICD.

123I-mIBG imaging might also help evaluate patients with primary arrhythmias. Mitrani et al128 observed that in patients who presented with VT but had structurally normal hearts, 55% had regional sympathetic denervation compared with none of the control patients. Gill et al129 found asymmetrical uptake of 123I-mIBG (less in septum) in 47% of patients with VT and “clinically normal” hearts, particularly obvious in patients with exercise-induced VT.

Interestingly, PET autonomic tracer (i.e., 11C-HED) imaging of particular primary arrhythmias, such as right ventricular outflow tract tachycardia130 and Brugada syndrome, shows focal defects in specific myocardial walls.131 Regional autonomic abnormalities can also be seen in nonischemic cardiomyopathies, such as Chagas disease in which the posterolateral, inferior, and apical walls are selectively affected. In one study of 26 patients with chronic Chagas cardiomyopathy, 123I-mIBG defects correlated with the occurrence of sustained VT.132 Further investigation of such observations may lead to better understanding of the pathophysiology of cardiac autonomic innervation, as well as the particular disease entities.

Ischemic Heart Disease

123I-mIBG imaging also shows promise in the setting of ischemic heart disease. Sympathetic fibers are more sensitive to ischemia than myocytes.133 MI causes sympathetic denervation beyond the infarcted area.56,134-137 Injury to sympathetic innervation may persist after myocyte recovery, resulting in areas of autonomic/perfusion mismatch, possibly pre-disposing to post-MI arrhythmias.57,138-140 Tomoda et al141 showed that 3-4 weeks after nonST segment elevation MI, 123I-mIBG defects may be present without Tl-201 perfusion defects. 123I-mIBG defects can occur following angina,142 and may be present up to 6 months after coronary spasm.143,144 Simula et al145 reported potential use in detecting subclinical disease, finding autonomic image abnormalities in asymptomatic patients with normal 99mTc-sestamibi studies but with left anterior descending disease. Subclinical endothelial dysfunction may cause episodes of vasoconstriction, resulting in neurohormonal event sequences that affect local sympathetic output.

123I-mIBG imaging can also shed light on the effect of sympathetic alteration on post-MI LV remodeling. Sakata et al146 found that after a first MI, despite a patent infarct coronary artery, the presence of a high severity score correlated with LV end-systolic volume dilatation.

Autonomic imaging is under investigation in the setting of “hibernating” myocardium. Using a porcine model, Luisi et al147-150 have produced large regional autonomic defects that increase in size and severity over time, increasing the likelihood of arrhythmic SCD. Similar abnormalities in sympathetic nerve function in chronic ischemic disease without infarction have been described in humans, such as Hartikainen et al151 finding regional 123I-mIBG defects in almost all patients with a >50% coronary stenosis. Among those with a stenosis >90%, 123I-mIBG defect size was indistinguishable from patients with previous MI.

To examine further the utility of autonomic imaging in hibernating myocardium, the Prediction of ARrhythmic Events with Positron Emission Tomography (PAREPET) trial has been undertaken. In this observational cohort study, >200 patients with ischemic cardiomyopathy (NYHA Class I-III CHF, EF ≤ 35%), without plans for coronary revascularization, underwent 13NH3 PET perfusion imaging, 18FDG myocardial viability imaging, and 11C-HED imaging.152 Preliminary data demonstrate significant variability in the extent of viable, dysinnervated myocardium, from small borders around areas of infarction to large confluent regions encompassing several myocardial segments.153 At the 2012 Heart Rhythm Society meetings, Fallavollita reported that the 4-year occurrence of sudden cardiac arrest (arrhythmic death or ICD shock for VT ≥ 240 minutes or ventricular fibrillation) increased in relation to the severity/extent of autonomic image abnormalities independent of BNP, CHF symptoms, or LVEF; of note autonomic/perfusion mismatch was not an independent predictor of adverse outcome.154 While this work was with a PET autonomic tracer, similar principles should apply to imaging with 123I-mIBG.

Autonomic Imaging in Other Conditions

Autonomic imaging has been shown to have potential use in other clinical conditions. Akutsu et al155 found that in HF patients with paroxysmal atrial fibrillation (AF), a decreased H/M was independently predictive of transition to permanent AF.

Following cardiac transplant reinnervation is important, and its absence may indicate cardiac pathology.156 Estorch et al157 studied patients 6 months to 12 years post-transplant, finding that H/M correlated with time after transplantation, indicating progressive reinnervation. Patients with absent tracer uptake were more likely to develop coronary vasculopathy.

Numerous reports have shown that in diabetic patients, uptake abnormalities of 123I-mIBG or PET autonomic tracers correlate with a worsened prognosis, even in the absence of clinical neuropathy.158,159 Nagamachi et al160 followed 144 patients without evidence of heart disease for 7.2 years, finding that a combination of decreased H/M and heart rate variability (HRV) abnormalities independently predicted events, and delayed H/M alone predicted all-cause mortality. Yufu et al161 reported that abnormal 123I-mIBG washout and age were independently associated with major cardiac and cerebrovascular events. Further work should determine if neuronal imaging in diabetics can effectively detect higher risk than is clinically apparent.

Given the enhanced sensitivity of sympathetic nerve to insults, some have investigated potential use of autonomic imaging to identify early myocardial damage from cancer chemotherapy. Olmos et al162 found decreased 123I-mIBG uptake as the cumulative dose of doxorubicin increased, with subsequent deterioration in LVEF. Carrió and colleagues found that at a cumulative doxorubicin dose of 240-300 mg/m2 123I-mIBG abnormalities correlated with 111In-antimyosin antibody uptake, although they found no association with LV functional impairment.163 Clinical utility beyond current monitoring methods needs further investigation.

Looking Forward

Prospective studies in larger study populations are required to establish the clinical utility of 123I-mIBG imaging in the various clinical scenarios discussed such that it is accepted by general cardiologists and allied physicians. It is important to demonstrate that the technique can effectively guide therapy to improve patient outcome and well being. Given that cardiac autonomic innervation is linked to underlying molecular and electrophysiologic processes of disease, radionuclide autonomic imaging promises to yield information that other imaging techniques cannot. Autonomic imaging may provide unexpected insights and understanding of cardiac diseases, and lead to new therapies. It is important that those in the field of nuclear cardiology encourage and participate in the development of this method.