Molecular Mechanisms of Adenosine Stress T1 Mapping

Abstract

Background:

Adenosine stress T1 mapping is an emerging MRI method to investigate coronary vascular function and myocardial ischemia without application of a contrast agent. Using gene-modified mice and two vasodilators, we elucidated and compared the mechanisms of adenosine myocardial perfusion imaging and adenosine T1 mapping.

Methods:

Wild-type (WT), adenosine A_2A (A_2AAR^−/−), A_2B (A_2BAR^−/−), and A₃ receptor knockout (A₃AR^−/−), and endothelial nitric oxide synthase receptor knockout (eNOS^−/−) mice underwent rest and stress perfusion MRI (n=8) and T1 mapping (n=10) using either adenosine, regadenoson (a selective A_2AAR agonist), or saline. Myocardial blood flow and T1 were computed from perfusion imaging and T1 mapping, respectively, at rest and stress to assess myocardial perfusion reserve (MPR) and T1 reactivity (ΔT1). Changes in heart rate (ΔHR) for each stress agent were also calculated. Two-way ANOVA was used to detect differences in each parameter between the different groups of mice.

Results:

MPR was significantly reduced only in A_2AAR^−/− compared to WT mice using adenosine (1.06±0.16 vs 2.03±0.52, p<0.05) and regadenoson (0.98±026 vs 2.13±0.75, p<0.05). In contrast, adenosine ΔT1 was reduced compared to WT mice (3.88±1.58) in both A_2AAR^−/− (1.63±1.32, p<0.05) and A_2BAR^−/− (1.55±1.35, p<0.05). Furthermore, adenosine ΔT1 was halved in eNOS^−/− (1.76±1.46, p<0.05) vs WT mice. Regadenoson ΔT1 was approximately half of adenosine ΔT1 in WT mice (1.97±1.50, p<0.05), and additionally, it was significantly reduced in eNOS^−/− mice (−0.22±1.46, p<0.05). Lastly, ΔHR was two times greater using regadenoson vs. adenosine in all groups except A_2AAR^−/−, where heart rate remained constant.

Conclusions:

The major findings are that: (a) while adenosine MPR is mediated through the A_2A receptor, adenosine ΔT1 is mediated through the A_2A and A_2B receptors, (b) adenosine MPR is endothelial-independent while adenosine ΔT1 is partially endothelial-dependent, and (c) ΔT1 mediated through the A_2A receptor is endothelial-dependent while ΔT1 mediated through the A_2B receptor is endothelial-independent.

Introduction

Adenosine stress T1 mapping is an emerging MRI method to interrogate coronary vascular function and myocardial ischemia without the use of contrast agents¹. In this method, adenosine-induced coronary vasodilation leads to an increase in myocardial blood volume (MBV), and since the T1 of blood is greater than the T1 of myocardium, the increase in MBV is reflected as an increase in the aggregate T1 of heart tissue, an effect referred to as adenosine T1 reactivity. Initial clinical evaluations have detected reduced adenosine T1 reactivity in patients with coronary microvascular disease due to type 2 diabetes² and in ischemic regions in patients with coronary artery disease (CAD)³. A study in dogs with acute coronary stenoses suggested that MBV may better differentiate between moderate and severe coronary stenoses than myocardial blood flow (MBF)⁴, and that it may be a more comprehensive global marker of ischemia as it represents the total blood volume of the coronary macro- and microcirculation^4–6. Also, adenosine T1 reactivity imaging may be particularly useful in patients with contraindications to gadolinium unable to undergo adenosine contrast-enhanced myocardial perfusion imaging, such as those with end stage renal disease⁷. While data suggest the potential clinical utility of adenosine T1 reactivity imaging, the molecular mechanisms underlying adenosine T1 reactivity have yet to be carefully elucidated, potentially limiting the ability to fully understand and interpret adenosine T1 reactivity findings.

Adenosine mediates vascular dilation, increased MBF, and increased MBV through interactions with cell surface receptors on coronary vascular smooth muscle cells (VSMCs) and endothelial cells (ECs). There are four adenosine receptor (AR) subtypes, namely A₁, A_2A, A_2B, and A₃, and all four are expressed in the coronary vasculature. The A_2AAR is expressed on VSMCs and ECs and is the subtype that predominantly controls MBF by relaxing arteriolar vascular smooth muscle, leading to decreased coronary resistance⁸. With these properties, the A_2AAR would be expected to play an important role in mediating MBV and adenosine T1 reactivity. The A_2BAR is also present on coronary VSMCs and ECs, and studies have shown its role in mediating MBF⁹, albeit with a much smaller effect than that of the A_2AAR, suggesting the possibility of a role for the A_2BAR in MBV and T1 reactivity. The A₃AR and A₁AR are also present on coronary VSMCs, and both have been shown, to a relatively small degree, to negatively modulate coronary vasodilation mediated by activation of other adenosine receptor subtypes^10,11, suggesting the potential for modest opposing effects on MBV and T1 reactivity. Finally, whether the vasodilatory effects of adenosine on MBF and MBV are endothelial-dependent or endothelial-independent remains incompletely understood, and elucidating these mechanisms would be important to the clinical interpretation of adenosine myocardial perfusion and T1-reactvity imaging studies.

Using gene-modified mice, adenosine or the selective A_2AAR agonist, regadenoson, and small-animal MRI for quantitative myocardial perfusion imaging and myocardial T1 mapping, we investigated the roles of the A_2AAR, the A_2BAR, and the A₃AR in both adenosine myocardial perfusion imaging and myocardial T1 reactivity. We also utilized endothelial nitric oxide synthase (eNOS) knockout mice to investigate whether adenosine myocardial perfusion and myocardial T1 reactivity are endothelial-dependent or -independent.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Experimental Design

We first sought to determine the contributions of the A_2A, A_2B, and A₃ ARs to the increases in myocardial perfusion and T1 after vasodilation using adenosine or regadenoson. For these studies, we performed myocardial perfusion MRI (n=8 per group) and T1 mapping (n=10 per group) before and 10 minutes after injection of vasodilator in 8–12-week-old male wild-type (WT), A_2AAR knockout (A_2AAR^−/−), A_2BAR knockout (A_2BAR^−/−), and A₃AR receptor knockout (A₃AR^−/−) mice, all on a C57Bl/6 background.

Next, we examined whether the changes in myocardial perfusion and T1 during vasodilation are attributable to endothelial-independent or dependent pathways. For these studies, we performed the same MRI protocol as above in 8–12-week-old male eNOS knockout (eNOS^−/−) mice to evaluate MBF and T1 reactivity.

Animal Handling

All animal studies were performed in accordance with protocols that conformed to the Declaration of Helsinki as well as the Guide for Care and Use of Laboratory Animals (NIH publication no. 85–23, revised 1996) and that were approved by the Animal Care and Use Committee at the University of Virginia. WT mice, A_2AAR^−/− mice (stock no. 010685), and eNOS^−/− mice (stock no. 002684) were obtained from The Jackson Laboratories (Bar Harbor, ME). A_2BAR^−/− mice and A₃AR^−/− mice were generated by Dr. Katya Ravid of Boston University (Boston, MA) and Merck Research Laboratories (West Point, PA), respectively, and were bred at the University of Virginia. During MRI, mice were anesthetized with 1.25% isoflurane and maintained at 36±0.5°C, and the electrocardiogram, body temperature, and respiration were monitored. An intraperitoneal (IP) catheter was established for delivery of vasodilatory agents.

MRI Acquisition and Analysis

MRI was performed using a 7T system (Clinscan, Bruker, Ettlingen, Germany) with a 35-mm diameter birdcage RF coil and an MR-compatible physiological monitoring and gating system for mice (SA Instruments, Inc., Stony Brook, NY). The MRI protocol began with localizer imaging to select a mid-ventricular short-axis slice. Next, as shown in Figure 1, T1 mapping and perfusion imaging were performed at rest. Thereafter, adenosine (Sigma–Aldrich, St. Louis, MO, 250 mg/kg/min¹², IP infusion), regadenoson (Sigma–Aldrich, St. Louis, MO, 0.1 mg/kg¹³, IP bolus), or saline was given and after 10 minutes, both T1 mapping and perfusion imaging were repeated. The total infusion or injection volume ranged from 80 to 100 μl. Heart rate (HR) measurements were taken before and 10 minutes after administration of the vasodilator. All image reconstructions and analyses were performed offline using MATLAB (Mathworks, Natick, MA). Regions of interest (ROIs) for the myocardium were drawn manually and included the entire left ventricular myocardial area within a slice, using conservative delineation of the endocardial and epicardial borders.

Figure 1.

Schematic timing diagram of the rest and stress perfusion and T1 mapping experiments. The output parameters were the myocardial perfusion reserve (MPR) and T1 reactivity (ΔT1).

Perfusion imaging was performed using a steady pulsed arterial spin labeling (ASL) sequence, a method previously shown to be insensitive to variations in HR¹⁴. Imaging parameters for ASL included: echo time/repetition time (TE/TR) = 2.5/10.0 ms, field of view (FOV) = 25 × 25mm², matrix size = 128 × 128, flip angle = 7°, slice thickness = 1 mm, saturation band thickness = 10 mm, number of averages = 9, and total scan time of 6 minutes. MBF was calculated as previously described¹⁴. MPR was calculated as: MPR = stress MBF / rest MBF.

T1 mapping was performed using a spiral Look-Locker sequence with cardio-respiratory gating, as previously described^15,16. The technique uses fuzzy-clustering of spiral k-space interleaves¹⁷ to ensure accurate T1 estimation, even with varying respiratory or heart rates. Using k_y-t undersampling with randomness and a low-rank reconstruction¹⁸, the acquisition was accelerated by a factor of two. Imaging parameters included: TR = 7 s, TE = 0.67 ms, flip angle = 3°, FOV = 30 × 30 mm², number of averages = 3, number of spiral interleaves = 84, interleaves per heartbeat = 3 and in-plane resolution = 0.23 × 0.23 mm². The total scan time was approximately seven minutes. To quantify myocardial T₁, signal intensity-time curves from myocardial ROIs were fitted to: $M_z(t)=M_0+(M_z(0)-M_0)\mathrm{e}\mathrm{x}\mathrm{p}(-t/T_1)$. An optimal fit of the model parameters M_0, M_z(0), and T₁ was obtained by minimizing the mean squared error. T1 reactivity was calculated as: ΔT1 = % change from rest to stress T1.

To assess the intersession repeatability of our MRI methods, ASL and T1 mapping were performed on 8 WT mice at rest and after adenosine vasodilation on two separate imaging sessions spaced one day apart. Bland-Altmann plots were used to analyze the discrepancy and the limits of agreement, and the intraclass correlation coefficient (ICC) and coefficient of variation (COV) were calculated for both measurements.

Statistical Analysis

Statistical analysis was performed using SigmaPlot (Systat Software Inc., Point Richmond, CA). Two-way ANOVA with a post-hoc Tukey’s HSD test was used to detect specific group-to-group differences in MBF, T1, MPR, ΔT1, and heart rate between different mouse strains and vasodilator types.

Results

ASL and T1 mapping repeatability

Bland-Altman plots showing the intersession repeatability of ASL and T1 mapping are shown in Figure 2. For ASL, the mean MBF difference was 0.1 ml/g/min and the 95% confidence interval was −3.0 to 3.2 ml/g/min. The ICC and COV for ASL were 0.90 and 20.4 ± 8.6 %. For the T1 mapping sequence, the mean difference was 0.4 and the 95% confidence interval was −43.9 to 44.7 ms. The ICC and COV for T1 mapping were 0.94 and 1.7 ± 0.1 %.

Figure 2.

Bland-Altman plots for the (A) ASL and (B) T1 mapping MRI sequences demonstrating small average differences and narrow limits of agreement between the two measurements of each MRI sequence.

Adenosine-induced increases in myocardial perfusion are mediated solely through the A_2AAR

Figure 3A shows example myocardial perfusion maps acquired at rest and during adenosine-induced vasodilation using ASL in WT and A_2AAR^−/− mice, demonstrating the effect of adenosine on increasing MBF in WT mice but not in A_2AAR^−/− mice. The results summarizing the pre- and post-adenosine, pre- and post-regadenoson, and pre- and post-saline perfusion imaging studies using WT, A_2AAR^−/−, A_2BAR^−/−, and A₃AR^−/− are presented in Table 1 and Figure 3B. In WT control mice, MBF at rest and with adenosine were 6.3 ± 2.3 and 12.1 ± 3.7 respectively, leading to a reference MPR value of 2.1 ± 0.5. In contrast, MPR due to adenosine was reduced to approximately unity in A_2AAR^−/− mice, demonstrating a central role of the A_2AAR in mediating the increase in MBF. Perfusion imaging of A_2BAR^−/− and A₃AR^−/− mice show that MBF at rest and MBF with adenosine are not significantly different than the results in WT mice (Table 1), and that MPR is not significantly different than that of WT mice (Figure 3B). MBF and MPR results measured using regadenoson in the various groups of mice were not significantly different than those measured using adenosine. These results indicate that the increase in perfusion due to either vasodilator is mediated through the A_2AAR.

Figure 3.

A. Example ASL perfusion maps before and after adenosine vasodilation in wild type (WT) and A_2AAR^−/− mice. B. Adenosine, regadenoson, and saline myocardial perfusion reserve (MPR) measurements for WT and adenosine receptor subtype knockout mice. MPR in response to adenosine and regadenoson was reduced in A_2AAR^−/− mice as compared to all other groups of mice.

* p<0.001vs WT adenosine and WT regadenoson

† p<0.001 vs adenosine in all other mouse groups

‡ p<0.001 vs regadenoson in all mouse groups

§ p<0.001 vs A_2BAR^−/− adenosine and A_2BAR^−/− regadenoson

‖ p<0.001 vs A₃AR^−/− adenosine and A₃AR^−/− regadenoson

Table 1.

Myocardial blood flow (ml*g⁻¹*min⁻¹) at rest and stress in response to adenosine, regadenoson, and saline in wild type (WT) and adenosine receptor subtype knockout mice (mean ± stdev).

	Adenosine		Regadenoson		Saline
	Rest	Stress	Rest	Stress	Rest	Stress
WT	6.3 ± 2.3	12.1 ± 3.7 *	6.2 ± 1.3	12.8 ± 3.6 *	6.3 ± 2.3	6.2 ± 2.2 ‡
A2AAR−/−	6.7 ± 2.1	7.3 ± 3.4 †	8.0 ± 1.6	7.9 ± 2.8 †	6.7 ± 2.1	6.7 ± 2.1
A2BAR−/−	7.2 ± 1.2	13.5 ± 2.5 *	7.3 ± 2.1	14.2 ± 3.7 *	7.2 ± 1.2	6.7 ± 0.9 ‡
A3AR−/−	7.8 ± 2.4	15.2 ± 4.9 *	7.0 ± 1.5	14.1 ± 2.2 *	7.8 ± 2.4	7.5 ± 2.3 ‡

^*p<0.001 vs same vasodilator rest perfusion in same group of mice

^†p<0.001 vs same vasodilator stress perfusion in all other groups of mice

^‡p<0.001 vs adenosine and regadenoson stress perfusion in same group of mice

Adenosine-induced myocardial T1 reactivity is mediated through both the A_2AAR and the A_2BAR

Example T1 relaxation curves computed from Look-Locker images acquired before and after adenosine vasodilation in a WT mouse and regadenoson vasodilation in an A_2AAR^−/− mouse are shown in Figure 4A and 4B. Table 2 and Figure 4C summarize the pre- and post- adenosine, regadenoson, and saline T1 and ΔT1 data from WT, A_2AAR^−/−, A_2BAR^−/−, and A₃AR^−/− mice. T1 reactivity due to adenosine is reduced by approximately a factor of 2 in both A_2AAR^-/ and A_2BAR^−/− mice compared to WT mice, while there is no significant difference in adenosine T1 reactivity between WT and A₃AR^−/− mice. These results demonstrate that A_2AAR and A_2BAR agonism each account for about half of adenosine T1 reactivity, and the A₃AR does not have a significant role in adenosine T1 reactivity. As shown in Table 2, we also observed that T1 values are lower in A_2BAR^−/− and A₃AR^−/− mice than in WT and A_2AAR^−/−mice.

Figure 4.

A. Example myocardial T1 relaxation curves before and after adenosine in a wild type (WT) mouse and (B) regadenoson in an A_2AAR^−/− mouse. C. Adenosine, regadenoson, and saline T1 reactivity (ΔT1) measurements in WT and adenosine receptor subtype knockout mice. Adenosine ΔT1 was significantly reduced in both A_2AAR^−/− and A_2BAR^−/− mice. Regadenoson ΔT1 was approximately half of adenosine ΔT1 in WT mice and was nullified in A_2AAR^−/− mice.

* p=0.030 vs WT adenosine

† p<0.001 vs WT adenosine and WT regadenoson

‡ p=0.045 vs WT adenosine, p=0.037 vs A₃AR^−/− adenosine

§ p=0.032 vs A_2AAR^−/− adenosine, p=0.039 vs WT regadenoson, p=0.044 vs A_2BAR^−/− regadenoson, p=0.048 A₃AR^−/− regadenoson

‖ p=0.027 vs A_2AAR^−/− adenosine

# p=0.044 vs WT adenosine, p=0.036 vs A₃AR^−/− adenosine

** p=0.031 vs A_2BAR^−/− adenosine, p=0.022 vs A_2BAR^−/− regadenoson

†† p=0.021 vs A₃AR^−/− adenosine

‡‡ p<0.001 vs A₃AR^−/− adenosine, p=0.050 vs A₃AR^−/− regadenoson

Table 2.

Myocardial T1 at rest and stress in response to adenosine, regadenoson, and saline in wild type (WT) and adenosine receptor subtype knockout mice (mean ± stdev).

	Adenosine		Regadenoson		Saline
	Rest	Stress	Rest	Stress	Rest	Stress
WT	1236 ± 43	1284 ± 41 *	1255 ± 69	1280 ± 79	1242 ± 55	1236 ± 58
A2AAR−/−	1213 ± 53	1232 ± 47	1254 ± 67	1252 ± 62	1253 ± 61	1251 ± 58
A2BAR−/−	1098 ± 65 †	1113 ± 70 †	1149 ± 37 †	1172 ± 45 †	1158 ± 44 †	1157 ± 58 †
A3AR−/−	1107 ± 72 †	1154 ± 64 ‡	1164 ± 45 †	1184 ± 45 †	1168 ± 75 †	1169 ± 64 †

^*p=0.026 vs WT adenosine rest T1

^†p<0.01 vs same treatment WT and A_2AAR^−/− mice

Adenosine, a nonselective adenosine receptor agonist, provides greater myocardial T1 reactivity compared to regadenoson, a selective A_2AAR agonist

Table 2 and Figure 4C also show that adenosine T1 reactivity in WT mice is about two times greater than T1 reactivity due to regadenoson. In A_2BAR^−/− mice, adenosine and regadenoson T1 reactivity are similar and approximately half of the value measured in WT mice. Regadenoson T1 reactivity is similar in WT, A_2BAR^−/− and A₃AR^−/− mice and is substantially reduced in A_2AAR^−/− mice. Further, the amount of regadenoson-induced T1 reactivity in all but the A_2AAR^−/− mice corresponds to about half of the adenosine-induced T1 reactivity measured in WT mice. These results demonstrate that adenosine agonizes both the A_2AAR and A_2BAR and that each contribute to T1 reactivity, whereas regadenoson agonizes only the A_2AAR, and thereby contributes to about half of the adenosine-induced T1 reactivity.

Adenosine- and regadenoson-induced increases in myocardial perfusion are eNOS-independent

In our second set of experiments we performed rest and vasodilator ASL and T1 mapping in eNOS^−/− mice. Table 3 and Figure 5A summarize the perfusion data before and after injection of each stress agent, showing no significant difference in MBF or MPR between WT and eNOS^−/− mice using either vasodilator, indicating that both adenosine- and regadenoson-induced increases in myocardial perfusion are eNOS independent.

Table 3.

Myocardial blood flow (ml*g⁻¹*min⁻¹) at rest and stress in response to adenosine, regadenoson, and saline in wild type (WT) and eNOS^−/− mice (mean ± stdev).

	Adenosine		Regadenoson		Saline
	Rest	Stress	Rest	Stress	Rest	Stress
WT	6.3 ± 2.3	12.1 ± 3.7 *	6.2 ± 1.3	12.8 ± 3.6 †	6.3 ± 2.3	6.2 ± 2.2 ‡
eNOS−/−	5.5 ± 1.0	10.7 ± 1.3 §	6.7 ± 4.3	13.9 ± 7.9	5.5 ± 1.0	5.3 ± 1.0 ‖

^*p=0.004 vs WT adenosine rest

^†p= 0.002 vs WT regadenoson rest

^‡p=0.003 vs WT adenosine stress, p<0.001 vs WT regadenoson stress

^§p<0.001 vs eNOS^−/− adenosine rest

^‖p<0.001 vs eNOS^−/− adenosine stress, p=0.03 vs eNOS^−/− regadenoson stress

Figure 5.

Adenosine, regadenoson, and saline (A) MPR and (B) ΔT1 measurements for wild type (WT) and eNOS^−/− mice. There was no difference in MPR between adenosine and regadenoson MPR in WT vs eNOS^−/− mice. However, adenosine ΔT1 was significantly lower in eNOS^−/− mice and regadenoson ΔT1 was nullified in eNOS^−/− mice.

* p=0.003 vs WT adenosine, p=0.001 vs WT regadenoson

† p=0.009 vs eNOS^−/− adenosine, p=0.001 vs eNOS^−/− regadenoson

‡ p=0.030 vs WT adenosine

§ p<0.001 vs WT adenosine, p=0.003 vs WT regadenoson

‖ p=0.014 vs WT adenosine

# p=0.036 vs eNOS^−/− adenosine, p=0.012 vs WT regadenoson

** p=0.028 vs eNOS^−/− adenosine

A_2AAR-mediated myocardial T1 reactivity is eNOS dependent while A_2BAR-mediated myocardial T1 reactivity is eNOS independent

Figure 5B and Table 4 show the myocardial T1 and ΔT1 results from WT and eNOS^−/− mice treated with adenosine, regadenoson, or saline. For adenosine T1 reactivity, eNOS^−/− mice have approximately half the ΔT1 response as WT mice. Using the selective A_2AAR agonist regadenoson, eNOS^−/− T1 reactivity was substantially reduced. Together, these results suggest that A_2AAR-mediated myocardial T1 reactivity is eNOS dependent and that A_2BAR-mediated myocardial T1 reactivity is eNOS independent. All T1 reactivity and absolute change in T1 values are shown in Table I and II in the Data Supplement.

Table 4.

Myocardial T1 at rest and stress in response to adenosine, regadenoson, and saline in wild type (WT) and eNOS^−/− mice (mean ± stdev).

	Adenosine		Regadenoson		Saline
	Rest	Stress	Rest	Stress	Rest	Stress
WT	1236 ± 43	1284 ± 41 *	1255 ± 69	1280 ± 79	1242 ± 55	1236 ± 58
eNOS−/−	1250 ± 83	1272 ± 92	1200 ± 52	1198 ± 56	1256 ± 72	1253 ± 68

^*p=0.04 vs WT adenosine rest

Heart rate responses to adenosine and regadenoson

Figure 6 summarizes the heart rate responses to each vasodilator for the different groups of mice. For both adenosine and regadenoson, the increases in HR were not significantly different for WT, A_2BAR^−/−, A₃AR^−/−, and eNOS^−/− mice, and changes in heart rate are minimal in A_2AAR^−/− mice. Regadenoson provides an approximately two-fold greater heart rate increase compared to adenosine in all groups except A_2AAR^−/− mice.

Figure 6.

Adenosine, regadenoson, and saline heart rate response post-administration in all groups of mice. The change in heart rate (ΔHR) with regadenoson was about twice that of adenosine in all groups of mice except A_2AAR^−/− where ΔHR was essentially eliminated.

* p=0.025 vs WT adenosine

† p=0.009 vs WT adenosine, p<0.001 vs WT regadenoson

‡ p=0.004 vs WT adenosine, p=0.007 vs A_2BAR^−/− adenosine, p=0.033 vs A₃AR^−/− adenosine, p=0.039 vs eNOS^−/− adenosine

§ p<0.001 vs regadenoson in all other mouse groups

‖ p=0.046 vs A_2BAR^−/− adenosine

# p=0.021 vs A_2BAR^−/− adenosine, p<0.001 vs A_2BAR^−/− regadenoson

** p=0.011 vs A₃AR^−/− adenosine

†† p=0.044 vs A₃AR^−/− adenosine, p<0.001 vs A₃AR^−/− regadenoson

‡‡ p<0.001 vs eNOS^−/− adenosine

§§ p=0.031 vs eNOS^−/− adenosine, p<0.001 vs eNOS^−/− regadenoson

Discussion

The major findings of this study, as show in Figure 7, are: (a) while adenosine myocardial perfusion reserve is mediated solely through the A_2AAR, adenosine myocardial T1 reactivity is mediated through both the A_2AAR and the A_2BAR, (b) while adenosine increases MBF through eNOS-independent mechanisms, adenosine increases MBV and T1 reactivity partly through eNOS-dependent mechanisms, and (c) specifically, the increase in MBV and T1 reactivity mediated through the A_2AAR is eNOS-dependent, while the increase in MBV and T1 reactivity mediated through the A_2BAR is eNOS-independent.

Figure 7.

(A,B) Five groups of mice underwent rest and stress perfusion and T1 mapping MRI using adenosine, regadenoson, or saline. The output parameters were myocardial perfusion reserve (MPR) and T1 reactivity (ΔT1). C. We found that adenosine increases myocardial blood flow (MBF) via endothelial-independent activation of the A_2AAR. In contrast, adenosine increases myocardial blood volume (MBV) and subsequently T1 via both endothelial-dependent activation of the A_2AAR and endothelial-independent activation of the A_2BAR.

The differential effects of the adenosine receptor subtypes and eNOS on MPR and T1 reactivity may be explained by considering the different sections of the coronary vasculature tree that primarily control MBF and MBV. Increased MPR requires primarily dilation of the resistance vessels, which are predominantly the coronary arterioles^19–22. Our results suggest that the increase in MBF secondary to vasodilation of the coronary arterioles is entirely A_2AAR-dependent and endothelial-independent, and these results are consistent with prior studies^19,23–26. Adenosine has also been implicated in capillary recruitment^27,28 and dilation²⁹, and as the terminal arterioles and capillaries with diameters <10μm contain the majority of MBV, this may be the main mechanism of myocardial adenosine T1 reactivity. Our results show an endothelial-dependent role of the A_2AAR and an endothelial-independent role of the A_2BAR in modulating MBV. The A_2AAR may directly stimulate eNOS^30–32 or may indirectly stimulate eNOS through flow-mediated dilation^33,34, whereas the A_2BAR may directly stimulate VMSCs on the terminal arteriole vessels^26,35 resulting in increased capillary recruitment and MBV. Flow-mediated dilation may also occur upstream of the coronary arterioles, however, the effect of the upstream vessels on MBV and T1 reactivity is likely overshadowed by the downstream terminal arterioles and capillaries. Overall, our results demonstrate that adenosine T1 reactivity imaging interrogates different aspects of coronary vascular function than adenosine perfusion imaging and suggests that some of the change in MBV occurs downstream of changes in MBF.

Our results in mice are not entirely consistent with published results of adenosine and regadenoson myocardial T1 reactivity measurements in human subjects. First, for MRI of the heart at 7T, our myocardial T1 values are low. While our fully-sampled spiral Look-Locker method provides accurate myocardial T1 values¹⁷, the accelerated version of the method causes an underestimation of T1³⁶. While not ideal, reproducibility remains high and the four-fold faster scan facilitates the acquisition of ASL and T1 mapping during a single adenosine stress session. We also report an adenosine T1 reactivity in WT mice of 3.9 ± 1.5%, while adenosine T1 reactivity in healthy humans is generally reported to be approximately 6%^1–3. Differences in T1 and T1 reactivity may be related to the use of different pulse sequences. For example, Kuijpers et al. reported an adenosine T1 reactivity of 4.3 ± 2.8% in patients using a breathhold modified Look-Locker inversion recovery (MOLLI) sequence as opposed to the 6.2 ± 1.0% reported using the standard breathhold shortened MOLLI (shMOLLI) sequence³⁷. For mouse MRI, we used a low-rank accelerated segmented Look-Locker acquisition method which may yield different T1 reactivity values than other sequences. Furthermore, as myocardial T1 is affected by changes in MBV, there may be inter-species differences in MBV between mice and humans that may affect T1 and T1 reactivity. Third, we report a significantly reduced regadenoson T1 reactivity vs. the T1 reactivity due to adenosine whereas Van Dijk et al. reported a regadenoson T1 reactivity of 5.4 ± 2.4% which was not significantly different than adenosine ΔT1 in healthy patients³⁸. A possible explanation for this discrepancy is that the distribution of adenosine receptors along the coronary artery tree may vary from species to species⁸, potentially altering the T1 reactivity mechanisms of mice compared to humans.

Prior studies have shown a small vasoconstrictive role for the A₃AR that occurs in response to A_2AAR and A_2BAR mediated vasodilation^11,39. Our results may be consistent with these studies as we observed a 10% increase in adenosine T1 reactivity in A₃AR^−/− compared to that of WT mice (although the increase was not statistically significant), potentially reflecting a small vasoconstrictive role for the A₃AR. Future studies may utilize IB-MECA, an A₃AR agonist, and CAY10498 or PSB11, A₃AR antagonists, in combination with gene-knockout mice to further elucidate a potential effect of the A₃AR on adenosine T1 reactivity.

In WT mice, our rest and stress perfusion measurements of 6.3 and 12.1 ml/g/min, respectively, and MPR measurement of 2.1 are consistent with prior studies^13–15. However, there is conflicting evidence regarding the role of the endothelium in the adenosine-induced increase in MBF. Multiple studies have suggested that adenosine imparts an endothelial-independent increase in MBF^19,23–26, however, PET perfusion studies have demonstrated a 21–25% reduction in stress MBF after inhibition of eNOS^40,41. While the role of the endothelium is controversial, our results add to the evidence that the adenosine-induced increase in MBF is endothelial-independent.

Prior studies have shown roles of the A_2AAR and A₁AR in modulating heart rate^8,42,43, and, in particular, Schindler el at demonstrated that A_2AAR activation in the central nervous system elevates heart rate while A₁AR activation in the peripheral nervous system decreases heart rate⁴³. Our results are consistent with these studies as regadenoson increased heart rate in mice about 60 beats per minute more than adenosine. Our results further suggest that A₁AR activation may act to suppress heart rate in parallel with an A_2AAR-mediated increase, as there was no change in heart rate in A2A^−/− mice due to adenosine. As a related point, we previously demonstrated that spiral Look-Locker MRI with cardio-respiratory gating and fuzzy clustering measures myocardial T1 independent of heart rate¹⁷. Heart rate independent T1 mapping was particularly important in this study because we required precise T1 mapping in a setting where different vasodilators applied in various gene-modified mice led to a variety of heart rates.

A limitation of our study is that we did not include A₁AR^−/− mice. Similar to the A₃AR, pharmacological studies have indicated that A₁AR activation may induce coronary vasoconstriction, and thus may negatively modulate T1 reactivity to a small degree^44,45. Future studies may investigate a potential role for the A₁AR in adenosine myocardial T1 reactivity. A second limitation is that our study was performed using male mice; however, there are sex-dependent differences in MBF and MBV and their response to adenosine⁴⁶. Thus, future studies will be required to determine whether these findings can be reproduced in female mice. Furthermore, there are sex differences in coronary vascular disease, including coronary macrovascular and microvascular disease^47–49. Whether our results hold or may be modified, potentially in a sex-dependent manner, in various disease contexts may also be investigated in future studies.

Clinical Perspective.

The results of the present study show that there are differences in the adenosine receptor subtypes that mediate adenosine myocardial perfusion imaging and adenosine T1 reactivity imaging. Adenosine myocardial perfusion is mediated solely through the A_2A adenosine receptor (AR), while adenosine T1 reactivity is mediated through both the A_2AAR and the A_2BAR. Furthermore, while adenosine increases myocardial perfusion through eNOS-independent mechanisms, adenosine increases myocardial blood volume and T1 reactivity partly through eNOS-dependent mechanisms. Due to these differences, while adenosine myocardial perfusion imaging and adenosine T1 reactivity imaging may both be used clinically for ischemia detection, they are not equivalent and should not be considered interchangeable in their clinical interpretation. The two techniques have fundamental differences in their underlying molecular mechanisms, and their clinical applications and the resulting clinical interpretation and performance may reflect these different underlying mechanisms.