Noninvasive assessment of coronary vasodilation using cardiovascular magnetic resonance in patients at high risk for coronary artery disease

Asymptomatic patients with diabetes mellitus (DM, N = 8) and end stage renal disease (ESRD, N = 5), as well as age-matched controls (N = 33) were recruited consecutively. Subjects were excluded if they had a history of chest pain, coronary artery disease, myocardial infarction, stroke or peripheral vascular disease. Vasoactive medications were discontinued 24 hours before the examination. All subjects provided written informed consent approved by the Human Subjects Committee at Stanford University.

A 1.5-T Signa MRI scanner (GE Healthcare, Milwaukee, Wisconsin) equipped with high-performance gradients (40 mT/m, 150 mt/m/ms) and a real-time interactive workstation were used. A commercial coil provided signal reception (5-inch General Purpose Coil, Model #2127316, GE Healthcare, Milwaukee, Wisconsin). Blood pressure and heart rate were monitored throughout the study (Omega 1400, In vivo Research, Inc., Orlando, Florida).

NTG-induced coronary vasodilation was performed as previously described [13]. A real-time interactive MR system [13‐15] was used to localize coronary arteries (16 frames/sec, TR 4.6 ms, flip 30, slice thickness 7 mm, FOV 24 cm, in-plane resolution 2.7 mm). High-resolution coronary MRA was then performed using a cardiac-gated, breath-held, multi-slice spiral sequence [13] (FOV = 20–28 cm, slice thickness = 5 mm, TR = 1 heartbeat, TE = 7 ms, in-plane spatial resolution = 0.62–0.99 mm, 14 to 20 interleaves, flip angle = 60 degrees, acquisition gated to diastole). In-plane and cross-sectional images were acquired before and then 5 minutes after 0.4 mg sublingual NTG, which was given while the subject was in the magnet. Images were reconstructed onto a 512 × 512 matrix, yielding a pixel size of 0.39 to 0.55 mm. Real-time short axis views of the left ventricle (LV) from the apex to base as well as 4-, 3- and 2-chamber views were also obtained to evaluate LV function.

For quantitative analysis of coronary vasodilation, the cross-sectional right coronary artery (RCA) images were used, except in subjects with a small non-dominant RCA, where the cross-sectional left anterior descending artery (LAD) images (n = 5) were used. As described previously[13] the slice with the most circular cross-section was identified on the pre-NTG images and the corresponding post-NTG slice was carefully matched according to the surrounding cardiac and chest wall structures. These images were then pooled and randomized, with no patient information or NTG status provided on the images, and then analyzed independently and in a blinded fashion by one observer. A custom designed software program was used to analyze the cross sectional images: after images were magnified two-fold, an ovoid region of interest tool was used to trace around the RCA or LAD, yielding the cross-sectional area (CSA). This analysis has been shown previously to have a low intra- and inter-observer variability [12, 13] and good correlation with x-ray coronary angiography [13].

Data were expressed as mean values ± standard deviation. The difference in coronary artery size before and after NTG was compared by a paired two-sided t test. Differences in % coronary vasodilation between patients and controls were tested using an unpaired two-sided t test. Differences among subject groups were tested using a one-way analysis of variance (ANOVA) and Fisher exact test. All statistical analyses were performed with StatView (version 5, SAS Institute Inc., Cary, NC).

Abnormal coronary vasomotor function occurs early in the development of atherosclerosis [16] and is typically assessed by studying endothelial-dependent epicardial coronary vasodilation by invasive methods [5, 17]. However, several x-ray coronary angiography studies [6, 10] have also found impairment of endothelial-independent coronary vasodilation to NTG in patients, with prognostic significance. In a study of 147 patients with risk factors for CAD (including 9% with diabetes and 84% with angiographic evidence of atherosclerosis) followed for a median 7.7 years, abnormal vasodilator response to acetycholine, cold pressor, and NTG were each independently associated with disease progression and increased cardiovascular events [6]. Consistent with these findings, a study of 163 women (including 26% with diabetes and 45% without angiographic evidence of CAD) found impaired reactivity to NTG and acetycholine in subjects who had future cardiovascular events [10].

Prospective data on coronary vasomotor function are limited in patients with DM [18] and ESRD [19]. However, consistent with an increased cardiovascular risk, peripheral vasomotor dysfunction has been almost universally found in these patients [20‐26]. Two studies in non-insulin-dependent DM have shown impaired brachial artery vasodilation to both endothelial-dependent and endothelial-independent stimuli [24, 26]. Interestingly, in a study using brachial ultrasound in subjects with no documented CAD (including 13.1% with DM), the only risk factor independently associated with impaired NTG-induced vasodilation was DM [27]. Similar to findings in the peripheral vasculature, reduced coronary artery reactivity to NTG has been previously reported in patients with DM. In a study [9] of non-insulin-dependent DM using intravascular ultrasound, coronary artery distensibility and diastolic cross sectional luminal area after NTG were significantly lower in DM compared to controls.

Similar to patients with DM, patients with renal failure have impaired peripheral vasomotor function [19, 28, 29]. In a study of 28 patients with chronic renal failure (13 on hemodialysis), both flow-mediated and NTG-induced brachial artery vasodilation were impaired to comparable degrees [28]. To date, there are no known studies on coronary vasoreactivity in patients with ESRD.

Our data show that patients with DM and ESRD have impaired NTG-induced coronary vasodilation compared to age-matched controls. The individual data (Figure 3b) reveal that patients generally fell into two groups: those with normal coronary vasodilation (~25%) and those with low coronary vasodilation (< 15%). Eighty percent (80%) of ESRD and 38% of DM patients fell below the 15% threshold, compared to only 6% of the controls.

CMR [30, 31], computed tomography (CT) [32, 33], ultrasound [34‐36], and nuclear techniques [37, 38] all offer alternative approaches to assess coronary artery disease noninvasively. By using sub-mm spatial resolution and analyzing the lumen cross-sectional area, CMR has been shown to have adequate resolution to detect coronary vasodilation to NTG in two prior studies [12, 13]. CMR can also directly image the coronary wall, with increased wall thickness demonstrated in patients with Type I DM [39] and non-obstructive CAD [39, 40]. CT can provide high-resolution structural imaging of the coronary lumen and wall [32], but the radiation and contrast involved make it suboptimal for serial imaging of coronary vasomotor changes. One recent study did look retrospectively at patients who had more than one coronary CT scan, where NTG was used in one and not in another, and did show significantly larger coronary diameter with NTG [41]. The feasibility of transthoracic echocardiography for measuring epicardial coronary vasodilation has recently been shown in healthy men [42]. The other main approach to assess coronary function noninvasively has been to measure coronary flow or perfusion reserve to a vasodilator stimulus (e.g., adenosine). This is primarily a measure of coronary microvascular function in the absence of epicardial stenoses. This can be performed by CMR [30, 31, 43], positron emission tomography (PET) [37, 38], and transthoracic Doppler techniques [34‐36] and has been shown to be impaired in patients with coronary risk factors [31, 38], including DM[35, 38]. More data comparing the prognostic significance of epicardial vs. microvascular vasomotor function are needed.

A major study limitation is that only endothelium-independent coronary vasodilation with NTG was evaluated. Endothelium-dependent vasomotor function has been shown to be an earlier marker of atherosclerosis and may be more sensitive in patients with subclinical disease [2, 4, 16]. Future studies should focus on overcoming the challenges of performing a more endothelial-dependent stimulus in the magnetic resonance environment. A preliminary report [44] and a case report [45] on using the cold pressor test show promise [44]. In addition, the size of the clinical cohort was small and recruitment was consecutive, which may have contributed to the significant differences in demographics (i.e., % female) between the two groups. This difference did not account for the finding of impaired vasodilation in the patients, as this finding remained significant even when the only males were analyzed. A study incorporating more female patients is needed to verify if the impaired coronary vasodilation applies to high-risk women. Finally, long-term clinical follow up is needed to determine the prognostic significance of these findings.