Here, we have demonstrated the ability to estimate perfusion during rest and exercise stress using systolic FAIR. We compared the use of 2RR interval delays between inversion pulse and imaging acquisition to 1RR delay and found that the former approach yielded higher TSNR at both rest and stress at the expense of producing, on average, 6% lower MBF values during rest.
Similar to previous studies exploring the use of 1RR and 2RR FAIR techniques, we found a reduction in MBF for the 2RR approach (29). This may be explained by the partial labeling of the inflowing blood for 2RR but not 1RR FAIR. Despite the loss of MBF accuracy it can be advantageous to use a 2RR delay as it appears to increase the TSNR, which is particularly important during stress when physiological noise is higher. Although the lower MBF for 2RR compared to 1RR FAIR was significantly different only during rest, we hypothesize that any similar effect during stress may be masked by the increase in MBF variability during this scan. This variability may be attributed to differences in respiratory and cardiac motion between the 1RR and 2RR FAIR scans which are typically larger during rest than stress, but also actual differences in cardiac workload between the 1RR and 2RR scans.
The rest mean MBF was approximately 1.2-1.4 mL/g/min in this study. Others also reported a similar range (1.3-1.5 mL/g/min) (10,13), but which is above the considered normal limit of 0.8 mL/g/min. The measured MBF was doubled during stress which is rather low considering the rule of thumb of 3.5 times increase (30). We observed only a modest heart rate increase in the scanned healthy subjects during exercise stress, which may indicate that the work load was relatively low, yielding a similarly moderate perfusion increase. Another source of reduced difference between rest and stress perfusion could be that the subjects were slightly anxious in rest. However, physiological stress derived from muscle work is not the same as pharmacological stress. Furthermore, no correction was made for how well-trained the people were, and fitness status may explain differences and the variability in physiological response to stress. The reduction in TSNR during stress compared to rest may be attributed to a combination of changes in the systolic rest period relative to rest (which would increase physiological noise) and actual changes in perfusion across the 6 breath-holds. Further work is required to limit the adverse effects of physiological noise during stress, including improved respiratory motion compensation such as prospective correction or breathing guidance. Physiological noise due to cardiac motion could be further reduced by shortening the acquisition window using alternative acceleration techniques. Finally, the use of pharmacological stress is likely to be extremely beneficial for reducing respiratory and cardiac motion (mainly due to ECG-mistriggering cause by the exercise) and will be explored in future patient studies. Nevertheless, the proposed systolic FAIR technique appears robust and may offer an attractive approach for myocardial perfusion assessment during stress without the use of contrast agents or pharmacological stress.
A clinical exercise test aims to reach an age-based heart rate, to induce ischemia which may be hidden at rest (31). In this study, a modest significant heart rate increase of 35% was achieved, which for patients might not be sufficient given the low average age of the subjects. However, similar experiments have been performed to investigate whether a difference in perfusion could be detected using hand grip exercise (13). The scanning protocol and equipment in this study was designed to evaluate a particular cardiac ASL technique in healthy subjects. To develop a clinical protocol requires either pharmacological stress where drugs are given strictly according to weight and tolerance or exercise with an age-based target heart rate (31), and including at least three short-axis views of the heart (4).
The clinically most common modality for perfusion measurement is Single Photon Emission Computed Tomography (SPECT), but it provides low temporal and spatial resolution. Photon Emission Tomography (PET) is considered the clinical gold standard but requires the use of ionizing radiation, unlike MRI (32). There are various MRI techniques used clinically to measure perfusion in the brain without Gd contrast, some of which have been implemented for cardiac MRI (8). In particular, ASL, blood oxygenation level dependent sequence (BOLD) (7), intravoxel incoherent motion (IVIM) (33) and T1 mapping (34) have been evaluated in animal experiments, healthy subjects and patients. ASL is the most widely used non-contrast media MRI technology in the brain due to its robustness (35) and therefore probably has the greatest chance of also succeeding in the heart.
The study has several limitations: The study comprised of a small number of healthy subjects, and larger studies including patients with coronary artery disease are warranted to evaluate this technique, including validation against quantitative reference techniques such as contrast-enhanced MRI. Furthermore, validation using a perfusion phantom would be desirable and will be the focus of future work (36,37). The use of a step ergometer to stress the myocardium has practical challenges, including a likely higher incidence of large-scale bulk respiratory and cardiac motion compared to conventional pharmacological stress where patients can lie still. Furthermore, the step ergometer requires additional patient preparation to ensure the equipment is tightly fastened to the patient which increases scan complexity and adds examination time overhead. However, it can be desirable to avoid the use of pharmacological stress agents, and alternative exercise approached may be explored in those patients (38). A technical limitation of this technique is the requirement for relatively long breath-holds which can be challenging to consistently maintain during stress, and particularly for patients with cardiovascular disease. Respiratory-induced motion in the through-plane direction may cause MBF quantification errors if it occurs between the slice-selective inversion and image acquisition, which may be particularly problematic for the 2RR FAIR technique where this delay is the longest. Prospective motion correction can mitigate these limitations, and techniques to enable motion-tolerant free-breathing FAIR would be desirable to facilitate patient scans and clinical translation (19,39,40).