Myocardial perfusion magnetic resonance imaging (MRI) with sliding-window conjugate-gradient highly constrained back-projection reconstruction (SW-CG-HYPR) allows whole left ventricular coverage, improved temporal and spatial resolution and signal/noise ratio, and reduced cardiac motion-related image artifacts. The accuracy of this technique for detecting coronary artery disease (CAD) has not been determined in a large number of patients. We prospectively evaluated the diagnostic performance of myocardial perfusion MRI with SW-CG-HYPR in patients with suspected CAD. A total of 50 consecutive patients who were scheduled for coronary angiography with suspected CAD underwent myocardial perfusion MRI with SW-CG-HYPR at 3.0 T. The perfusion defects were interpreted qualitatively by 2 blinded observers and were correlated with x-ray angiographic stenoses ≥50%. The prevalence of CAD was 56%. In the per-patient analysis, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of SW-CG-HYPR was 96% (95% confidence interval 82% to 100%), 82% (95% confidence interval 60% to 95%), 87% (95% confidence interval 70% to 96%), 95% (95% confidence interval 74% to100%), and 90% (95% confidence interval 82% to 98%), respectively. In the per-vessel analysis, the corresponding values were 98% (95% confidence interval 91% to 100%), 89% (95% confidence interval 80% to 94%), 86% (95% confidence interval 76% to 93%), 99% (95% confidence interval 93% to 100%), and 93% (95% confidence interval 89% to 97%), respectively. In conclusion, myocardial perfusion MRI using SW-CG-HYPR allows whole left ventricular coverage and high resolution and has high diagnostic accuracy in patients with suspected CAD.
Considerable progress has been made in myocardial perfusion magnetic resonance imaging (MRI) since its conceptualization in the early 1990s. However, conventional single-shot data acquisition method is limited by the low spatial coverage, temporal and spatial resolution, signal/noise ratio (SNR), and cardiac motion-related image artifacts. Accelerated imaging methods, such as generalized autocalibrating partially parallel acquisitions, sensitivity encoding, and echoplanar imaging can reduce the scan time per slice; however, the spatial coverage is still limited to 3 to 4 slices with relatively low resolution and SNR. Sliding-window conjugate-gradient highly constrained back-projection reconstruction (SW-CG-HYPR) allows whole left ventricular coverage, improved temporal and spatial resolution, and SNR, and reduced cardiac motion-related image artifacts. However, the accuracy of myocardial perfusion MRI with SW-CG-HYPR for detecting coronary artery disease (CAD) has not been determined in a large number of patients. Consequently, we conducted a prospective study to investigate the diagnostic value of this method for the detection of significant CAD.
Methods
The institutional research ethics committee approved our study. All patients gave written informed consent. From April 2009 to October 2010, a total of 77 consecutive subjects were recruited from patients with suspected CAD who were scheduled for primary diagnostic invasive coronary angiography. The exclusion criteria were medical instability, known CAD, any contraindications to MRI (i.e., metallic implants, such as pacemakers, defibrillators, cerebral aneurysm clips, ocular metallic deposits, severe claustrophobia) or contraindications to adenosine (e.g., second- or third-degree atrioventricular block, history of asthma), and renal insufficiency (i.e., estimated glomerular filtration rate assessed by creatinine clearance <60 ml/min/1.73 m 2 ). A total of 27 patients were excluded by these exclusion criteria. Thus, the final study population included 50 patients (28 men and 22 women; mean age 56 ± 16 years). All patients were instructed to refrain from any beverage or foods containing caffeine and from antianginal medication within 24 hours before the MRI study. The patients underwent MRI in the 2 weeks before coronary angiography. No clinical cardiac events were reported between the examinations.
All patients were examined with a 3.0 T whole-body scanner (Magnetom Trio, A Tim System, Siemens AG Healthcare, Erlangen, Germany) and a 12-element matrix coil. The R-wave acquired from a 4-electrode vector electrocardiogram was used to trigger the data acquisition. All data were acquired during end-inspiration to maximize the breath-hold capacity of the patients. The patients were instructed to hold their breath as long as possible during imaging and to resume shallow breathing when they could no longer hold their breath.
After localizer sequences, adenosine was administered intravenously at 140 μg/kg/min with the heart rate and blood pressure monitored. After 3 minutes of adenosine infusion, an intravenous bolus injection of 0.05 mmol/kg gadolinium contrast material (Magnevist, Schering, Berlin, Germany) was administered into an antecubital vein on the opposing arm using a power injector (Spectris, Medrad, Indianola, Pennsylvania) at a rate of 4 ml/s, followed by a 20-ml saline flush at 4 ml/s. A modified radial sampled fast low-angled shot sequence was used for image acquisition. The image parameters were as follows: repetition time/echo time/flip angle 2.92 ms/1.67 ms/12°, field of view 300 × 300 mm 2 , matrix 192 × 192, spatial resolution 1.6 × 1.6 × 10 mm 3 , TI 60 and 100 ms, and number of slices 6. One full k -space data set was acquired in an interleaved fashion with 16 projections per heartbeat for 10 cardiac cycles, and 6 data sets were collected for 60 heartbeats. Sliding composite images were reconstructed from k -space lines for 10 cardiac cycles. Time-resolved, contrast-enhanced images were reconstructed after CG-HYPR processing. A breath-hold fast low-angled shot sequence was used for cine imaging between the stress and rest perfusion imaging. The following sequence parameters were used: repetition time/echo time/flip angle 5.00 ms/2.44 ms/13°. The short-axis views were distributed to cover the entire left ventricle, and 2 long-axis views (4- and 2-chamber views) were also acquired. Ten minutes after stress perfusion imaging (the time in which the cine images were obtained), the rest perfusion images were acquired using the same imaging sequence without adenosine. Subsequently, another dose of 0.05 mmol/kg gadolinium contrast agent was administered immediately after the rest perfusion imaging for a total of 0.15 mmol/kg. Ten minutes later, a breath-hold segmented inversion-recovery prepared gradient echocardiographic sequence was used to assess myocardial viability. The total examination time was about 40 minutes.
Conventional x-ray coronary angiography was performed in all patients and quantitatively evaluated by coronary angiography (QuantCor QCA, Siemens AG Healthcare) by 2 independent blinded cardiologists in consensus. At least 2 orthogonal projections were evaluated. The degree of stenosis was evaluated in the projection showing the greatest degree of narrowing. A reduction in the luminal diameter of ≥50% in a major epicardial coronary artery or the major branches (≥2 mm) was considered relevant stenosis.
The data sets acquired using the SW-CG-HYPR sequence were reconstructed off-line using MATLAB (MathWorks, Natick, Massachusetts). The images were independently assessed by 2 experienced readers who were unaware of the clinical information. Any disagreement in the diagnosis between the 2 readers was settled by a consensus reading.
The readers were presented with anonymous MRI cases, including perfusion at stress, at rest, and delayed enhancement (DE). The image quality was graded on a 4-point scale: 1, poor with severe image artifacts; 2, moderate with moderate image artifacts; 3, good with minor artifacts; and 4, excellent with no apparent artifacts. Perfusion defects were assessed qualitatively. The presence of ischemia was defined by the following criteria: (1) territories without DE showing perfusion deficits during stress perfusion, but not at rest (stress-inducible deficits), for ≥3 consecutive image frames, and (2) territories with nontransmural DE demonstrating additional stress-inducible perfusion deficits. The perfusion and DE analysis of each myocardial segment was performed using the 16-segment model recommended by the American Heart Association. The 2 middle slices were combined to yield segments 7 to 12 (midcavity slice) of the model.
Statistical analysis was performed using the Statistical Package of Social Sciences, version 13.0 (SPSS, Chicago, Illinois). Continuous variables are expressed as the mean ± SD. Categorical variables are expressed as the absolute number (percentage). Student’s t test was used to assess the significance of continuous variables. κ Values were calculated to compare the interobserver agreement on a per-patient basis for myocardial territories. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were calculated according to standard definitions. Statistical tests were 2-tailed, and p <0.05 indicated a significant difference.
Results
All acquisitions were successful, and all images acquired were of sufficient quality for the final analysis. The mean image quality score of SW-CG-HYPR was 3.5 ± 0.5. The clinical details of the study group are summarized in Table 1 . X-ray coronary angiography demonstrated significant coronary artery stenoses in 56% of the patients (28 of 50). Six patients had 1-vessel disease, 9 patients had 2-vessel disease, and 13 patients had 3-vessel disease.
| Characteristic | Value |
|---|---|
| Age (years) | |
| Mean ± SD | 56 ± 16 |
| Range | 39–75 |
| Gender (n) | |
| Male | 28 |
| Female | 22 |
| Hypertension (≥140/90 mm Hg) | 18 (36%) |
| Smoker | 23 (46%) |
| Hypercholesterolemia (>220 mg/dl) | 27 (54%) |
| Diabetes mellitus | 9 (18%) |
| Family history of coronary artery disease | 16 (32%) |
| Number of coronary arteries narrowed ≥50% | |
| 1 | 6 (12%) |
| 2 | 9 (18%) |
| 3 | 13 (26%) |
| 0 | 22 (44%) |
Of the 50 patients, 45 (90%) experienced side effects during the adenosine infusion (breathlessness, flushing, headache), but no clinically relevant complications occurred. A significant decrease occurred in both systolic and diastolic blood pressure, with a significant increase in heart rate during stress ( Table 2 ).
| Variable | At Rest | Stress | p Value |
|---|---|---|---|
| Pulse rate (beats/min) | 67 ± 11 | 79 ± 13 | <0.05 |
| Systolic blood pressure (mm Hg) | 139 ± 21 | 124 ± 19 | <0.05 |
| Diastolic blood pressure (mm Hg) | 79 ± 10 | 72 ± 12 | <0.05 |
SW-CG-HYPR myocardial perfusion imaging correctly identified significant CAD in 27 of 28 patients (sensitivity 96%) and correctly ruled out CAD in 18 of 22 patients (specificity 82%; Figure 1 ) . SW-CG-HYPR did not detect perfusion defects in only 1 patient, who had 50% to 60% stenosis of the left anterior descending coronary artery. In 4 patients, SW-CG-HYPR detected perfusion defects despite normal coronary angiographic findings.
In the per-vessel analysis, the respective sensitivity and specificity for the left anterior descending artery was 96% (25 of 26) and 92% (22 of 24). For the left circumflex coronary artery, the corresponding values were 100% (15 of 15) and 97% (34 of 35), and for the right coronary artery, 100% (22 of 22) and 75% (21 of 28). The respective positive and negative predictive value for left anterior descending was 93% and 96%; left circumflex coronary artery, 94% and 100%; and right coronary artery, 76% and 100%. The sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy for the detection of relevant coronary stenosis for all coronary arterial vessels was 98%, 89%, 86%, 99%, and 93%, respectively.
A detailed overview of the diagnostic performance of myocardial perfusion MRI with SW-CG-HYPR for detecting CAD is summarized in Table 3 . Myocardial DE was visualized in 6 of the 50 patients suspected of having relevant CAD with no history of myocardial infarction and no electrocardiographic changes suggestive of myocardial infarction ( Figure 2 ) .
