The number of cardiac patients with chronic kidney disease is on the rise. Assessment of ischemic heart disease in these patients requires a non-contrast-enhanced, high-resolution, imaging approaches to evaluate the presence of perfusion anomalies. Myocardial BOLD MRI has the capacity to fill that unmet need. Over the past two decades myocardial BOLD MRI has seen major technical advancements and a number of clinical validation studies. However, the reliability of BOLD MRI remains a key weakness for its widespread adoption for routine clinical use. To date, most of the technical developments have addressed improvements in imaging speed, coverage and reducing image artifacts at rest. However, image artifacts from unpredictable cardiac motion during stress can lead to significant deterioration of image quality, which can confound/mask the BOLD signal changes during stress. Since the FDA approval in 2008, Regadenoson (Gilead Sciences Inc) has become a popular stress agent and is currently used in ~70%(1) of the pharmacological stress in the US owing to its improving patient tolerability and ease of administration. Studies have also shown that regadenoson can prolong the coronary vasodilation to 8-10 minutes (1), but to date no studies have quantified the myocardial blood flow (MBF) during the late phase of vasodilation. We hypothesized that (a) stress BOLD MRI performed at ~10 minutes can markedly improve the reliability for detecting myocardial hyperemia; and (b) that myocardial perfusion reserve remains significantly greater than 2.0 (a meaningful hyperemic state for ischemic testing) following regadenoson injection. We studied this using a canine model in a hybrid PET/MR system, which is capable of reporting BOLD signal changes (from MRI) and quantitative blood flow changes (from PET) in the heart. Healthy mongrel dogs (n=7) were studied in a state-of-the-art PET-MR system (Biograph mMR, Siemens Healthcare, Germany). After scouting and whole-heart shimming, BOLD images were acquired with 2D T2 maps at rest simultaneously with dynamic 13N-ammonia PET. Following a 40-min gap to ensure sufficient decay of radiotracer, a bolus injection of regadenoson (2.5 μg/kg) was administrated. T2 maps were acquired 2 mins and 10 mins post regadenoson administration (p.r.a) to investigate the mean BOLD response and reliability of the BOLD response. To quantify the extent of myocardial hyperemia 10 mins p.r.a, another dynamic 13N-ammonia PET was also acquired. Mean and standard deviation (s) of myocardial T2 were measured from images acquired at at rest and at 2 mins and 10 mins p.r.a.. Myocardial BOLD Response (indexed as T2(stress)/T2(rest)) and Myocardial BOLD Variability (indexed as sT2(stress)/sT2(rest)) and were computed at 2 mins and 10 mins p.r.a, respectively to assess mean BOLD response and the reliability of BOLD response. MR-based attenuation corrected PET images were analyzed in standard fashion with commercially available qPET software and were matched to the corresponding BOLD imaging slices to determine myocardial perfusion reserve (MPR) at 10 minutes p.r.a and regressed against Myocardial BOLD Response.Comparison of observed Myocardial BOLD Variability at rest, 2-mins and 10-mins p.r.a, along with representative corresponding mid-ventricular, short-axis, T2 maps are shown in Fig. 1. Note the extensive image artifacts present in the representative T2 map at 2 min (from the intense heart-rate variability during acquisition), which is absent in the T2 maps acquired at rest and 10-mins p.r.a.. Myocardial BOLD Variability was significantly larger at 2-min p.r.a (1.6±0.9) compared to 10-mins p.r.a (1.0±0.3) and at rest (1.0); p<0.05 for both. Average reduction in Myocardial BOLD Variability between 2-min and 10-mins p.r.a was 0.6±1.2. Representative 13N-ammonia PET images of MBF at rest and 10-mins p.r.a are shown in Fig. 2A. MBF at 10 min (1.8±0.9 ml/g/min) was significantly higher than at rest (0.6±0.3ml/g/min), p<0.05 (Fig. 2B). Mean MPR from PET at 10-min p.r.a was significantly larger than 2.0 (3.0±0.6, p<0.05). BOLD images corresponding to the PET images in Fig. 2A are shown in Fig. 2C. Myocardial T2 at 10-min p.r.a (40.4±1.7ms) was significantly higher than at rest (37.1±2.0ms), p<0.05 (Fig. 2D). Mean Myocardial BOLD Response at 10-min p.r.a was significantly higher than 1.0 (1.09±0.04<0.05). Strong correlation between PET MPR and Myocardial BOLD Response was observed (R=0.7, p<0.05); Refer to Fig. 2EMyocardial BOLD images acquired at 10-mins p.r.a were free of image artifacts typically observed in images acquired at 2-min p.r.a. MPR values at 10-mins p.r.a were significantly higher than 2.0 and were strongly correlated with the Myocardial BOLD Response. These data suggest that delayed BOLD acquisition following regadenoson administration can be a practical strategy for increasing the reliability of cardiac BOLD MRI for cardiac stress testing. The clinical utility of this approach remains to be evaluated in human subjects.