Awhile being a fundamental advantage of cardiac PET, absolute quantification is one of the major limitations of MRI. MR signal intensities depend on a vast array of parameters such as imaging hardware (transmitters, receivers, amplifiers, gradients, coils, etc.), pulse sequences, as well as many other features which vary between vendors, magnetic field strengths, and scanner models (Fig. 3.4). Although usually irrelevant in morphological imaging, quantification of MR signal is an important step in dynamic MR imaging. Due to its superior time resolution, the true potential in quantitative MRI lies in dynamic data, even if this term is used somewhat differently as compared to nuclear medicine. Basically, it refers to a series of images, where either acquisition parameters varied (such as the inversion time in T1 mapping or the gradients in diffusion imaging) or – more familiar to nuclear cardiology – which are acquired after the injection of MR contrast media.

Perfusion imaging is, as mentioned earlier, a traditional domain of nuclear imaging for the diagnosis of a flow-limiting stenosis in coronary artery disease (CAD), and the investigation of the hemodynamic significance of known CAD is of high clinical relevance. There is increasing evidence that PET MPI offers advantages over the traditional SPECT-based approach due to the absolute quantification of myocardial blood flow (MBF), the validated attenuation correction, and the superior image quality in obese patients [34, 35]. Three PET perfusion tracers have been established: ¹³N ammonia (¹³NH₃), ¹⁵O water, and rubidium (⁸²Rb). ¹³NH₃, ¹⁵O water, and ⁸²Rb have the advantage of a relatively short half-life, which is about 10 min for ¹³NH₃, 122 s for ¹⁵O water, and 76 s for ⁸²Rb, which allows for rapid PET MPI scans in rest and under stress. Recently, ¹⁸F flurpiridaz was introduced in early trials and has been shown to offer some advantages, e.g. a low positron energy and a half-life with the same logistical advantages as ¹⁸F-FDG [36].

But despite the established role of PET MPI, myocardial perfusion imaging using dynamic contrast-enhanced (DCE-) MRI is a research since almost two decades and has shown promising results when compared to PET [5, 37, 38]. MRI perfusion imaging has a much greater availability than cardiac PET, a lower examination cost, and no ionizing radiation. MRI usually offers much greater spatial in-plane resolution than PET, enabling the possibility of voxel-wise quantitative perfusion analysis with a resolution of at least 2 × 2 × 8mm³ – although still at the price of limited LV coverage [39]. A spatial resolution of three to six slices per cardiac cycle can be achieved with reasonable SNR, surpassing the temporal capabilities of PET imaging. Independent of the competitive ambitions of MRI to take its share in the large market of diagnosing CAD, there is also great potential for integrated systems concerning the creation of synergies. To elucidate that potential, we briefly compare “perfusion” imaging as realized by both modalities.

For PET, the frame rate is usually in the order of 10 s, which is too low to actually observe the tracer’s transit through the vascular space. Thus, PET MBF is estimated by observing tracer extraction from the vascular space into those parts of the myocardium where tracer kinetics are slow compared to the minimal frame rate. Retainable tracers as the ones listed above (with the exception of ¹⁵O water) are indeed conceptually quite similar to microspheres [40], however adapted for noninvasive application. The main design criterion for those compounds is how efficiently they are extracted from the blood pool and transferred into the myocytes. Depending on the tracer, multi-compartmental modeling is used to map the diffusive extraction from the blood pool and subsequent retention inside the myocytes to simpler rate constants. From the specific model, a first-pass extraction flow constant (often called K₁) is calculated, which can be directly identified with MBF for freely diffusible tracers (¹⁵O) or have to be corrected in case of high flow rates (¹³NH₃, ⁸²Rb). In contrast, the underlying principle of myocardial MRI perfusion imaging is the direct observation of indicator convection through the vascular space, enabled by the high temporal resolution of MRI. The mathematical framework for that is the classic indicator-dilution theory as adapted for noninvasive applications by Zierler [41, 42]. This formalism, unlike compartment modeling, does not necessarily impose a priori physiological information about the structure of the tissue. Perfusion flow is described directly as the delivery amplitude of contrast media from the LV lumen (where the AIF is acquired) to the distribution volume of contrast agent in the myocardial tissue. Although rarely applied in myocardial perfusion studies, extraction-based approaches more similar to PET, e.g. the Tofts model [43], are still widely used in cerebral and oncologic studies. It is important to note though that for highly vascularized tissue or high MBFs, the “extraction” of Gd-DTPA from the vascular space is permeability limited and not flow limited [44]. This renders it virtually impossible to extrapolate flow characteristics from Gd-DTPA extraction rates at high flows, which is why an extraction-based approach is usually not suitable for myocardial perfusion imaging at stress.

An additional, highly relevant point concerning MRI perfusion is that gadolinium-based contrast agents distribute in blood plasma, but do not permeate red blood cells (and thus the whole vascular space) as the more diffusible PET tracers do which clearly affects the generated signals (Fig. 3.5). As long as flow patterns of plasma and blood cells are homogeneous, this has no influence on time-related parameters like mean indicator transit times (MTT). However, there is a need to correct for the effective difference between large-vessel hematocrit and microvascular hematocrit [45] when estimating MBF. This might affect especially stress studies [46, 47].

To which extent this has consequences in a clinical setup is unclear yet but might be one of the reasons for the per-patient variability seen in comparative studies [5]. There are several additional factors stemming from the need to accurately estimate contrast agent concentrations from MRI signal values, making data post processing much more complicated in PET. In a similar fashion as for hematocrit, those factors may affect input functions and tissue curves differently, e.g. water exchange conditions [48], Gd-DTPA relaxivity [49], AIF dispersion [50], and blood flow effects. In conclusion, despite important differences in data structure to PET, new opportunities arise from the spatially and temporally highly resolved assessment of microvascular dynamics from MRI-based flow assessments. Especially for integrated PET/MRI systems, such synergistic approaches hold great promise for future clinical applications in multimodal perfusion imaging.

Targeting the water and its properties is another technique modulating MRI to retrieve “molecular” information. In oncological imaging, using advanced imaging sequences and strong gradients, this avenue was already explored: the apparent diffusion coefficient (ADC) of water [51] is already widely used to characterize tumorous tissue. In general, water is the predominant source of the MR signal and its motility (or diffusion properties) can be used for tissue characterization. Another approach is the delineation of the longitudinal relaxation time T1. This parameter describes the potential of the exited spins to transfer their energy to the surrounding macromolecules. Thus, one of the changes in the microchemical cellular environment was the introduction of MR medical imaging as T1 values differ between healthy and tumorous tissue [52]. The so-called T1 weighting (and its modification, inversion recovery imaging) is still a fundamental imaging sequence in diagnosis imaging as different T1 values will produce different signal intensities, and thus, the contrast between healthy and abnormal tissue is optimized. For inversion recovery imaging, any signal from remote myocardium is nulled and scar tissue is enhanced in late gadolinium enhancement (LGE) [53]. More specifically, it shows areas with a different distribution volume for MR contrast agents. Accordingly, different washout kinetics from the interstitial space was shown previously in the acute as in the chronic setting of myocardial infarction [54, 55]. Unfortunately, larger clinical validations are still lacking, but two general modes of T1 mapping in the heart are already used in an increasing number of MRI centers. In the last decade, novel cardiac MR technology allowed the calculation of pixel-wise T1 maps [56]. Native T1 maps were used in imaging regional edema, fibrosis, amyloidosis, Anderson-Fabry disease, intramyocardial hemorrhage, and other structural alterations, where a modified water content and protein environment in the myocardium occurs [6]. In addition, using MR contrast media, pre- and post-contrast T1 maps enable the calculation of the extracellular volume (ECV). It was already shown that the expansion of extracellular matrix has prognostic value [57] and might have an important role in heart failure imaging. Even the term “vulnerable interstitium” was already coined in sudden death patients [58]. Although this is speculative, cardiac PET/MRI might be an excellent platform to validate and improve the characterization of myocardial tissue by integrating this information with tracer signals.

Myocardial hibernation is a well-known phenomenon where myocytes shift their metabolism from fatty acids toward glucose after chronic hypoperfusion [59]. Several preclinical and clinical studies have proved the beneficial effects of revascularization which is followed by recovery of the contractile function [60]. Several modalities showed potential to identify hibernating myocardium: dobutamine stress echocardiography, MRI, ¹⁸F-FDG PET, and late gadolinium enhancement (LGE) MRI.

Considered as “gold standard” by many imagers, nuclear cardiology offers one of the most commonly applied techniques using ¹⁸F-FDG PET as marker of myocardial viability [61]. Typically, ¹⁸F-FDG is combined with a flow marker (¹³N-ammonia, ⁸²Rb, and also ^99mTc-labeled SPECT agents). Based on meta-analyses the sensitivity of ¹⁸F-FDG PET to predict functional recovery after revascularization is about 92 %, while the specificity is only 63 % [62]. However, there is evidence that patients with a significant amount of dysfunctional but viable tissue should undergo revascularization as soon as possible as viability is a transient state of the myocytes [63, 64]. Especially from the point of view of spatial resolution, however, MRI offers with delayed gadolinium enhancement (LGE) a readily available approach. MRI spatial resolution is superior to PET (1–3 mm vs. 4–6 mm) which allows the differentiation between transmural and non-transmural infarction [65]. The same advantage helps with a reduced or absent ¹⁸F-FDG uptake in thinned myocardium which would potentially be miscategorized as scar by PET. Despite these fundamental differences of the approaches (imaging viable vs. nonviable tissue), a good agreement between both modalities has been demonstrated in a chronic setting [53]. From a practical perspective, LGE is independent of the excretion of insulin and shows advantage in diabetic patients where the clearance of ¹⁸F-FDG from the blood is limited and cardiac image quality is poor. However, if a hybrid use of ¹⁸F-FDG PET MRI and LGE MRI is beneficial is still subject of research and only limited data is available indicating again good agreement (Fig. 3.6) [20].

In our initial experience in a similar setting, we found a small proportion of myocardial segments in disconcordance between ¹⁸F-FDG uptake and transmurality of LGE [66]. However, the clinical relevance for functional recovery in the long-term setting post-acute MI is debatable and warrants further research.

In the following paragraphs the delineation of (mostly) inflammatory signals is described in our first experience in distinct disease entities. Recent research has focused on the myocardial response to ischemia – with an emphasis on the cardiac inflammatory response after acute MI which is considered a risk factor for deleterious remodeling and finally the development of heart failure. The fact that monocytes migrate rapidly into the myocardium identified them as key components of cardiac inflammation [67–69]. This is a highly complex process and so far is the domain of preclinical research. However, ¹⁸F-FDG PET – due to its readily availability – is frequently used in a variety of settings, as activated inflammatory cells show increased expression of glucose transporters which can be visualized using ¹⁸F-FDG PET. This signal could be supplemented with MRI’s capability to image scar formation, microvascular obstruction, edema, hemorrhage, and left ventricular function. A recent preclinical study using a mouse infarct model of myocardial infarction showed that ¹⁸F-FDG PET/MRI quantifies the amount of accumulated monocytes in the postischemic heart if the glucose uptake of cardiomyocytes is suppressed [70]. Such a strategy would enable a detailed morphologic characterization of the infarcted heart by PET/MRI and is feasible in humans (Fig. 3.7) [71, 72].

In patients, the glucose metabolism of cardiomyocytes can be suppressed by a high-fat/low-carb diet (the so-called Atkins diet) and/or the administration of unfractionated heparin prior to ¹⁸F-FDG injection which is used, e.g., to image sarcoidosis with PET. However, there are limitations as the ¹⁸F-FDG uptake is not unique in the heart: as used in conventional viability imaging, ischemically compromised myocardium switches its metabolism from fatty acids toward glucose degradation [73]. In addition, expression of GLUT1 transporters has been demonstrated in postischemic myocardium and can persist over weeks and is responsible for the “ischemic memory” [74, 75]. Thus, ¹⁸F-FDG uptake in postischemic myocardium may not solely reflect inflammation but viable cardiomyocytes. However, if this characterization of tissue and its inflammatory state proves to be clinically reliable, it could identify patients at risk for developing heart failure or – even more relevant – it may serve as a tool to guide immunomodulatory therapeutic strategies.