Cardiac PET is the modality of choice to measure perfusion, metabolism and regional/absolute myocardial blood flow. PET imaging is facilitated by the use of radiotracers, in which a compound or pharmaceutical is labeled with a positron emitting radionuclide. Positrons are positively charged nuclear particles with the same mass as an electron. When an emitted positron collides with an electron in adjacent tissue, two 511 keV gamma rays (photons) are emitted in opposite directions. PET detectors are designed to register only photon pairs that strike opposing detectors at the same time, termed ‘coincidence detection’. PET radionuclides generally have much shorter half-lives than those used in single photon emission CT (SPECT), and therefore are associated with less radiation exposure [5].

The most commonly utilized cardiac PET tracers include: tracers to assess myocardial blood flow (MBF) (¹³N–Ammonia, Rubidium-82 [⁸²Rb], oxygen-15 labeled water [H₂ ¹⁵O]) or glucose metabolism (¹⁸F–Fluorodeoxyglucose, FDG). The type of image that can be acquired is based upon the pharmacokinetics of the particular radiopharmaceutical, and in principle, any physiological or molecular process that can be targeted with a radiotracer can be imaged, given suitable uptake in the area of interest. An important area of translational imaging research is the clinical development of specific molecular probes, which can detect inflammation, thrombosis, apoptosis, necrosis, angiogenesis, fibrosis or other alterations of the extracellular matrix [6].

Measurement of absolute MBF by PET entails the intravenous injection of a MBF tracer followed by dynamic acquisition of images of the radiotracer passing through the circulatory system and evaluation of extraction and retention in the myocardium. Kinetic models and operational equations (accounting for tracer decay, partial volume, blood pool spillover) are then applied to yield regional MBFs in absolute terms, ml/g/min [7]. Calculating the ratio of peak MBF at maximal pharmacologically induced vasodilatation to resting MBF provides quantification of myocardial perfusion reserve, which provides important prognostic and severity information in the assessment of coronary stenosis and microvascular disease [8‐10].

¹⁸F–FDG is a glucose analog that is taken up by myocytes (and other cells, including immune cells) via membrane glucose transporters (GLUT); and therefore highly sensitive to metabolically active processes. ¹⁸F–FDG is phosphorylated intracellularly in the glycolytic pathway by a hexokinase into ¹⁸F–FDG-6-phosphate. However ¹⁸F–FDG-6-phosphate cannot be metabolized further along this pathway and therefore accumulates within cells in direct proportion to their metabolic activity. In the myocardium, under aerobic conditions, most myocardial energy consumption is derived from oxidation of free fatty acids, followed by glucose, and therefore accumulation of ¹⁸F–FDG-6-phosphate is expected to be low, but increased in areas of increased metabolism, such as inflammation, or in the setting of myocardial hibernation.

CMR is an excellent non-invasive imaging modality for characterization of the heart as a complex structure and mechanical pump. Cine imaging is the reference standard for the assessment of structure and global ventricular performance through quantification of chamber volumes and function (ejection fraction (EF) and cardiac output). Combined with phase contrast imaging, valve regurgitation volume/fraction can also be reliably estimated for all four cardiac valves. Cine imaging can also be used to demonstrate regional wall motion abnormalities or LV thrombus. Newer forms of acquisition can directly assess myocardial deformation or strain (e.g. Tagging, Tissue phase mapping or DENSE imaging) [11, 12]. Assessment of diastolic function can also be performed with combinations of the above techniques, with characterization of atrial volume and mitral valve inflow patterns. In serial follow-up of ventricular function, CMR offers markedly superior inter-study reproducibility compared to 2D echocardiography [13].

CMR can also be used to characterize myocardial tissues based upon three properties that determine image contrast in an MR image: proton density (density of hydrogen atoms) and two magnetic relaxation parameters (T1 – longitudinal relaxation time) and T2 (Transverse relaxation time). For example, in T1-weighted images, myocardial tissue is dark while fat is bright. By contrast, T2-weighted images can be used to identify myocardial water/edema caused by inflammation or acute ischemia. Using a gadolinium-based contrast, which acts to shorten T1 in the area of distribution, can further assist tissue characterization. Gadolinium accumulates in areas of increased extracellular volume such as scar or fibrosis, resulting in hyperenhancement on inversion recovery CMR images – late gadolinium enhancement (LGE). It is possible, therefore, using a combination of acquisitions, which differ in terms of tissue contrast, to characterize normal from abnormal areas of myocardium and identify possible pathological mechanisms (e.g. fibrosis, edema, ischemia). Gadolinium CMR may also be used to quantify myocardial perfusion and perfusion reserve (in a conceptually similar way to PET) without radiation exposure using dual bolus, first pass imaging with Fermi function deconvolution methods [14, 15].

It is immediately clear that combining PET and CMR has great potential for the characterization of pathology from molecular to organ levels; this multimodality approach may benefit investigation of patients with heart failure.

Coronary PET-CMR imaging using ¹⁸F–FDG and ¹⁸F–sodium fluoride radiotracers in combination have been used recently to image coronary artery inflammation and microcalcification, respectively. This very new approach, has the potential to be used to identify patients with increased or active atherosclerotic disease activity that may benefit from aggressive risk factor modification [38].

Sarcoidosis is a systemic granulomatous disease in which inflammatory granulomas result in marked local inflammation and post-inflammatory scar. Cardiac granulomas produce conduction defects, heart block and heart failure [39]. The onset of cardiac involvement heralds a markedly worse prognosis for affected patients [40]. Early diagnosis and intervention with disease modifying therapy is therefore warranted. Historically Gallium-67 scintigraphy has been used for the diagnosis of cardiac sarcoid, but it has a low sensitivity owing to low image resolution, and has begun to be replaced by other imaging modalities such as CMR or PET. CMR has the potential to demonstrate myocardial edema using T2-weighted imaging, and also delineate scarring with LGE imaging, which has prognostic significance [41]. Importantly, inflammation is an important component of the disease process, and ¹⁸F–FDG PET may provide unique data on disease activity. Ishimaru et al. [42] showed abnormalities in PET imaging that were entirely absent on traditional Gallium scans. The complementary nature of CMR and PET imaging modalities has been investigated by Ohira et al. [43] who studied 21 patients with cardiac sarcoid using serial imaging techniques performed within 4 weeks. In this cohort of patients, the authors found focal ¹⁸F–FDG uptake in 15 patients (71%), and high signal intensity (T2-weighted imaging) or LGE on CMR in 9 patients (43%). Importantly, in eight of these patients, 3 had abnormal findings in only one of the two modalities, while the other five had positive findings on both images, but with different distributions. These findings are important as they indicate the potential power of orthogonal data from multimodality imaging to increase test sensitivity. This kind of data has been replicated in numerous case studies using hybrid or same day PET-CMR [44‐49]. Together these data indicate that the hybrid approach yields superior diagnostic information than either modality alone.

Cardiac amyloid deposits can be imaged using PET radiotracers, which target the amyloid β-pleated-sheet structure, for example ¹⁸F–florbetaben PET (Figs. 2, 50] CMR biomarkers of cardiac amyloidosis are of proven prognostic value [51]. However, CMR is unable to differentiate between the two main forms of amyloid: acquired monoclonal immunoglobulin light-chain (AL) and transthyretin-related (familial and wild-type/senile) amyloid (ATTR). This information has important prognostic and therapeutic implications. Nuclear imaging with SPECT bone tracers has been shown as one possible method to differentiate amyloid subtypes [52, 53]. Trivieri et al. [54] used hybrid PET-CMR and the PET bone tracer ¹⁸F–sodium fluoride, with the aim of differentiation of ATTR and AL forms within a single, low-radiation scan. Fourteen subjects were prospectively recruited, comprising 4 patients with biopsy-proven ATTR, 3 with biopsy-proven AL amyloid, and 7 control subjects without clinical suspicion of amyloid. The authors performed image analysis on fused, co-registered PET and MR LGE images, allowing PET activity to be measured within the LV myocardium, and importantly demonstrating specific areas of amyloid deposition that were visualized on LGE. The amyloid burden as assessed by T1 mapping was similar in patients with AL and ATTR amyloid, however patchy areas of increased ¹⁸F–sodium fluoride uptake were observed in the myocardium of patients with ATTR but not the other two groups. This study differs from some of the other approaches described in this review, in that truly independent (and complementary) data is derived from both modalities, rather than images which essentially describe the same pathology in different image space (e.g. ¹⁸F–FDG-PET and T1/T2 mapping for inflammation).

Novel PET radiotracers may have potential to further refine ATTR amyloid diagnosis. Amongst patients with ATTR V30 M (the most common Transthyretin [TTR] mutation in Europe) amyloidosis, two main phenotypes are observed: one characterized by late onset (>50 years) with both cardiac and neurological manifestations and another with an early onset of disease, with mainly neuropathic manifestations. These phenotypes have been associated with differences in amyloid fibril composition – type A (full length TTR and fragments) and type B (full length TTR) [55]. Pilebro et al. investigated the role of ¹¹C–Pittsburgh compound B (PIB) to differentiate these subtypes of ATTR, and showed that global LV ¹¹C–PIB was higher in patients with Type B V30 M ATTR [56].

Acute myocarditis is an important cause of death in children and young adults and is traditionally assumed to account for many cases of dilated cardiomyopathy previously classified as idiopathic. The role of CMR (especially T1 and T2 mapping) is now established in the diagnosis of acute and chronic myocarditis [57, 58]. However, CMR fails to identify a significant number of patients with biopsy proven myocarditis. The addition of nuclear imaging to CMR may therefore improve sensitivity, Fig. 3.

Nensa et al. investigated the feasibility of PET-CMR for the investigation of suspected myocarditis [59]. The authors showed good agreement between myocardial FDG and LGE/T2 hyper-intensity. Overall 24/65 patients had abnormalities on either PET or CMR imaging consistent with myocarditis: 17 patients (71%) showed concordant CMR and PET abnormalities, 6 patients (25%) had pathologic CMR findings in the absence of pathologic FDG uptake and 1 patient (4%) demonstrated pathologic FDG uptake in the absence of pathologic CMR findings. These data point to several important problems for PET-CMR: 1) in 8/65 patients (12%), the inhibition of physiological myocardial glucose uptake failed, rendering the nuclear imaging unusable, and 2) adding PET imaging provided only a minimal increase in the sensitivity above CMR imaging alone (also assuming this was not a false positive). On this basis, the case has not been clearly made for routine use of PET-CMR in the assessment of myocarditis.

Anderson Fabry disease (AFD) is an x-linked lysosomal storage disorder. It is caused by deficient activity of the lysosomal enzyme alpha-galactosidase A, resulting in the accumulation of globotriaosylceramide in lysosomes in multiple cell types throughout the body [60]. Heart failure is the most common first cardiac event in AFD patients [61], but may be preventable with enzyme replacement therapy; [62‐64] early diagnosis of cardiac involvement is therefore important. Nappi et al. [65] used PET-CMR hybrid imaging to assess cardiac involvement a cohort of 13 asymptomatic AFD patients. There were three components to the imaging protocol: LGE imaging (fibrosis), short inversion time inversion recovery [STIR] imaging (edema), and ¹⁸F–FDG PET (myocardial metabolism). 6/13 patients exhibited focal LGE indicating intra-myocardial fibrosis; four of these patients also had positive STIR imaging suggesting myocardial edema. All 4 patients with LGE and positive STIR showed focal FDG uptake in the corresponding myocardial segments indicating inflammation (these patients also had elevated cardiac troponin). An important area requiring further exploration is the significance of heterogenous FDG uptake in a group of patients without other CMR abnormalities - Does this finding represent myocardial inflammation indicating early (pre-clinical) cardiac involvement? Alternatively, could enzyme replacement therapy itself result in inflammation secondary to globotriaosylceramide clearance? This novel data supports the role of inflammation in the pathogenesis of cardiac involvement in AFD, and demonstrates the possibility for dissecting disease stages using multimodality PET-CMR.

Whilst PET-CMR literature remains in its infancy, it is possible to identify some areas of future development applicable to heart failure. Sympathetic nervous system imaging The ADMIRE-HF (AdreView Myocardial Imaging for Risk Evaluation in Heart Failure) study prospectively evaluated iodine-123 meta-iodobenzylguanidine (¹²³I–mIBG) imaging for identifying cardiac events in patients with symptomatic heart failure. Increased myocardial sympathetic activity is a prominent feature of heart failure and is associated with reduced neuronal norepinephrine (NE) uptake due to post-transcriptional down-regulation of the cardiac NE transporter, which can be assessed by ¹²³I–mIBG imaging. The ADMIRE-HF trial showed that mIBG data provides complementary prognostic information to existing heart failure biomarkers (BNP and LVEF) [66]. The role of the sympathetic nervous system can also be approached using PET tracers such as ¹¹C meta-hydroxyephedrine (¹¹C–HED) [67]. Data from ¹²³I–mIBG and ¹¹C–HED in heart failure have been shown to be analogous, with ¹¹C–HED having superior capacity to identify regional abnormalities [68]. The PAREPET trial (Prediction of ARrhythmic Events with Positron Emission Tomography) investigated sympathetic denervation assessed using ¹¹C–HED PET in ischemic cardiomyopathy. The study showed that ¹¹C–HED PET predicted cause-specific mortality from sudden cardiac arrest independently of LVEF and infarct volume. These data may provide information useful for the identification of patients most likely to benefit from an ICD [69]. The potential of multiparametric assessment of ischemic cardiomyopathy investigated by De Haan et al. [70] used serial PET and CMR with ¹¹C–HED (sympathetic innervation) and ¹⁵O–water (myocardial perfusion) to study the relationship between, scar, perfusion and innervation. The authors showed that areas of denervated, residual viable myocardium in ischemic cardiomyopathy were related to the heterogenic scar zone as assessed with LGE CMR. A novel ¹⁸F–labeled ligand for the norepinephrine transporter (N-[3-bromo-4-(3-¹⁸F-fluoro-propoxy)-benzyl]-guanidine [LMI1195]) is in clinical development for mapping cardiac nerve terminals in vivo using PET. First in man studies, have shown that LMI1195 is well tolerated and yields a radiation dose comparable to that of other commonly used PET radiopharmaceuticals [71]. The clinical and imaging significance of the autonomic nervous system for heart failure continues to develop [72].

The failing heart undergoes significant, complex changes in energy utilization. In addition to Glucose imaging with FDG-PET, it is also possible to assess myocardial fatty acid utilization using novel metabolic tracers such as ¹⁸F–Fluoro-6-Thia-Heptadecanoic acid (FTHA). Taylor et al. [73] demonstrated the feasibility of this approach, demonstrating a reciprocal relationship between myocardial fatty acid and glucose uptake rates in heart failure; the former were higher than expected for the normal heart, whereas myocardial glucose uptake rates were lower. Metabolic assessment using MRI approaches are also becoming feasible, through the use of hyperpolarized carbon-13 imaging of pyruvate metabolism. In a preclinical model of porcine cardiomyopathy, Schroeder et al. [74] successfully demonstrated impaired pyruvate and fat oxidation at the onset of cardiomyopathy. However, whether myocardial metabolism represents a useful biomarker for heart failure remains under uncertain.

Apoptosis is considered an important pathophysiological element to the development of ventricular dysfunction in heart failure. Several molecular markers of programmed cell death exist, which are attractive targets for molecular imaging – for example phosphatidylserine (PS), which is translocated from the inner plasma layer to the cell surface under conditions of cell stress and physiologically serves as a marker to macrophages to remove apoptotic cells [75]. Several ligands with high specificity for PS exist, including Annexin V, which has been studied in heart failure using technetium SPECT, although experimental PET radiotracers also exist [76, 77]. Targeted magnetic nanoparticles for use in MRI, have also been used to image apoptosis in experimental animal models [78].