|Home | About | Journals | Submit | Contact Us | Français|
We sought to evaluate prospectively the effects of heart rate and heart-rate variability on dual-source computed tomographic coronary image quality in patients whose heart rates were high, and to determine retrospectively the accuracy of dual-source computed tomographic diagnosis of coronary artery stenosis in the same patients.
We compared image quality and diagnostic accuracy in 40 patients whose heart rates exceeded 70 beats/min with the same data in 40 patients whose heart rates were 70 beats/min or slower. In both groups, we analyzed 1,133 coronary arterial segments. Five hundred forty-five segments (97.7%) in low-heart-rate patients and 539 segments (93.7%) in high-heart-rate patients were of diagnostic image quality. We considered P < 0.05 to be statistically significant. No statistically significant differences between the groups were found in diagnostic-image quality scores of total segments or of any coronary artery, nor were any significant differences found between the groups in the accurate diagnosis of angiographically significant stenosis.
Calcification was the chief factor that affected diagnostic accuracy. In high-heart-rate patients, heart-rate variability was significantly related to the diagnostic image quality of all segments (P = 0.001) and of the left circumflex coronary artery (P = 0.016). Heart-rate variability of more than 5 beats/min most strongly contributed to an inability to evaluate segments in both groups. When heart rates rose, the optimal reconstruction window shifted from diastole to systole.
The image quality of dual-source computed tomographic coronary angiography at high heart rates enables sufficient diagnosis of stenosis, although variability of heart rates significantly deteriorates image quality.
During the past few years, noninvasive coronary angiography upon multidetector-row computed tomography (MDCT) has rapidly progressed and has shown promise with regard to the detection and quantification of coronary artery stenosis.1–4 However, despite the increase in temporal resolution from 16- to 64-detector-row computed tomography (CT), coronary CT angiography remains sensitive to motion artifacts, which occur especially at higher heart rates.2,5–7 Results of a study3 of 64-detector-row CT coronary angiography showed a nonsignificant tendency toward lower image quality at higher mean heart rates, and a significant negative relation between image quality and heart-rate variability. In order to reduce motion artifacts, it has been proposed that patients be administered oral β-blocker medication for heart-rate control, even when 64-detector-row CT is to be used.8–11 In most studies that have involved 16- or 64-detector-row CT, the target for scanning has been maintained at heart rates slower than 70 or even 60 beats/min, so that good-quality images of coronary arteries could be obtained. The requirement to premedicate patients with β-blocker drugs in order to achieve a sufficiently low heart rate for scanning has been considered a major limitation surrounding the clinical use of MDCT coronary angiography.
Dual-source CT (DSCT) coronary angiography incorporates 2 X-ray tubes and 2 detectors that are mounted onto a rotating gantry, with an angular offset of 90°.12 The DSCT system affords a high temporal resolution of 83 ms in monosegment reconstruction mode. In contrast with single-source CT systems that rely on multisegment reconstruction techniques, temporal resolution upon DSCT is independent of heart rate. Initial studies have shown that DSCT enables the study of coronary arteries with excellent diagnostic quality in all patients, independent of heart rate—thus obviating the need to premedicate patients with β-blockers.12–15 We believed that the effects of heart rate and heart-rate variability on image quality, diagnostic accuracy, and optimal reconstruction windows merited further evaluation in patients whose heart rates exceeded 70 beats/min.
The aim of this study was to evaluate prospectively the effect of heart rate and heart-rate variability on DSCT image quality in patients who had high heart rates, and to determine retrospectively the accuracy of DSCT in the diagnosis of coronary artery stenosis, using invasive coronary angiography as the reference standard.
We enrolled 80 patients (53 men and 27 women; mean age, 49.9 ± 6.2 yr; age range, 38–71 yr) as our study group. All underwent contrast-enhanced cardiac DSCT. The patients were classified into 2 groups on the basis of their heart rates: 70 beats/min or slower (group 1, n=40), and 71 beats/min or faster (group 2, n=40). Of the 80 patients, 27 (19 men, 8 women) also underwent invasive coronary angiography. The mean time between the DSCT and conventional angiographic examinations was 22 days (range, 3–41 days). Thirty-one of the 80 patients (39%) were taking β-blocking agents as part of their baseline medication at the time of CT. No additional β-blockers were administered before the performance of CT. Referral for CT was on the basis of clinical indications only. Exclusion criteria were renal insufficiency (creatinine level, >120 μmol/L), known allergy to iodinated contrast material, hemodynamic instability, and unstable clinical condition. Patients who had coronary stents or who had undergone previous coronary artery bypass surgery were not included in this study. The study protocol was approved by our local ethics committee, and all patients provided their written informed consent.
The CT examinations were performed on a DSCT scanner (SOMATOM® Definition™ AS, Siemens Medical Solutions AG; Forchheim, Germany), with features as described above. Before undergoing scanning, all patients whose heart rates were slower than 100 beats/min received 1 mg of glycerol trinitrate sublingually. Nonenhanced DSCT for calcium scoring was performed, from 1 cm below the level of the tracheal bifurcation to the diaphragm, in a cranio-caudal direction. Each coronary angiographic scan was started upon continuous injection of a bolus of 65 to 85 mL of iopamidol (Iopamiro 370 mg/mL, Bracco S.p.A.; Milan, Italy), followed by the injection of 50 mL of saline solution into an antecubital vein through an 18-gauge catheter (injection rate, 5–5.5 mL/sec). Contrast-agent application was controlled by bolus tracking. A region of interest (mean diameter, 10.1 ± 3.4 mm; range, 7.5–15 mm) was established in the aortic root, and image acquisition started 5 sec after signal attenuation reached the predefined threshold of 100 Hounsfield units. The CT settings were as follows: detector collimation, 2 × 64 × 0.6 mm; pitch, adapted to heart rate (range, 0.2–0.46); rotation time, 330 ms; tube current, 400 mA-s per rotation; and tube voltage, 120 kV. Tube-current modulation was used to reduce the dose of current to the patients. Full current was applied from 35% to 70% of the R–R electrocardiographic interval.
By use of semiautomated software (syngo Calcium Scoring, Siemens Medical Solutions), a mean Agatston score was calculated for each patient from the nonenhanced DSCT data, with a detection threshold of 130 Hounsfield units. All images were reconstructed by means of retrospective electrocardiographic gating. A monosegment reconstruction algorithm, consisting of the data from a quarter rotation of both detectors, was used for image reconstruction. For each patient, data were initially reconstructed starting at 75% of the R–R interval with a slice thickness of 0.75 mm, a reconstruction increment of 0.5 mm, and a medium soft-tissue convolution kernel (B26f). If motion artifacts were present, additional reconstructions were performed in 5% decrements and increments. All images were transferred to an external workstation (LEONARDO, Siemens Medical Solutions) that was equipped with cardiac postprocessing software (syngo Circulation, Siemens Medical Solutions). Contrast-enhanced DSCT was evaluated by use of transverse source images and curved-planar reformation, along with thin-slab maximum-intensity projection and 3-dimensional volume rendering.
For data analysis, the coronary arteries were classified into 15 segments, in accordance with the model suggested by the American Heart Association.16 The right coronary artery (RCA) comprised segments 1 through 4, the left coronary artery (LCA, including the left main trunk and the left anterior descending coronary artery) comprised segments 5 through 10, and the left circumflex coronary artery (LCx) comprised segments 11 through 15. All segments with a diameter of at least 1.5 mm at their origin were included.2 All reconstructed images were independently analyzed and graded by 2 cardiovascular radiologists (MZ and HZ) who were blinded to heart rates and heart-rate variability. For each coronary segment, both readers inspected all optimal reconstruction windows, noting the presence of motion artifacts and evaluating image quality semiquantitatively in accordance with a previously described 4-point ranking scale.17 Image-quality scores for the arterial segments are shown in Table I. Lower scores corresponded with better image quality; diagnostic image quality was deemed achieved when the score was 3 or lower. Examples of the image-quality scoring are presented in Figure 1. Upon any disagreement in data analysis, discussion followed and consensus prevailed. Heart-rate variability was calculated as the SD from the mean heart rate.3
Conventional coronary angiography was performed in 27 patients (12 in group 1, 15 in group 2) in accordance with standard techniques, and at least 2 views in different planes were obtained for each coronary artery. All invasive coronary angiograms were evaluated by a cardiologist (JL) who was blinded to the CT results. Segmental disease was analyzed in each vessel via the same 15-segment model that had been used for DSCT analysis. Lesions constituting a stenosis that was 50% or more of a segment's diameter were deemed angiographically significant. Stenosis severity was classified on the projections with maximal-luminal-diameter stenosis. To visualize the RCA, at least 2 projections were obtained; for the LCA, at least 6 projections were obtained.
Statistical analysis was performed with commercially available statistical software (SPSS version 12.0 for Windows, SPSS Inc.; Chicago, Ill). Quantitative variables were expressed as mean ± SD. The Student t test was used to compare the differences between the groups. Interobserver agreement for image-quality assessment was interpreted by use of k values. Because of small samples in 2 groups, the diagnostic performance of DSCT in the detection of significant stenosis was presented as diagnostic accuracy only. Comparison between DSCT and conventional coronary angiography was performed on a per-segment basis. The effects of heart rate and heart-rate variability on image quality were evaluated by means of a Pearson correlation coefficient analysis for all coronary segments and for every vessel (RCA, LCA, and LCx) in both groups. A P value < 0.05 was considered to be statistically significant.
Mean heart rates were 62.5 ± 7.3 beats/min in group 1 and 79.9 ± 8.3 beats/min in group 2 (P < 0.001). There were no statistically significant differences between the groups in heart-rate variability or in distributions of age, sex, or body mass index. The characteristics of the patient population are listed in Table II. Mean heart-rate variability was 4.8 ± 2.2 beats/min in group 1 and 4.6 ± 2.8 beats/min in group 2. Twenty-two patients (27.5%; 14 in group 1, 8 in group 2) had stable heart rates that varied by only 1 or 2 beats/min during the entire examination. Seventeen patients (21.3%; 9 in group 1, 8 in group 2) had irregular heartbeats. The heart rates of these 17 patients were completely irregular during data acquisition, with rapid changes of more than 25 beats/min during the CT examination.
A total of 1, 133 arterial segments with diameters of at least 1.5 mm were evaluated in the 80 patients (21 segments were missing because of anatomic variations, and 46 segments had a diameter of less than 1.5 mm at their origin). The mean Agatston score was 464 ± 305 (range, 0–1, 867) in group 1, and 428 ± 312 (range, 0–1, 673) in group 2. Of 558 segments in group 1, images without motion artifacts (score 1) were obtained in 198 segments (35.5%); images with minor artifacts (score 2), in 275 segments (49.3%); and images with moderate artifacts (score 3), in 72 segments (12.9%). Severe artifacts (score 4) occurred in 13 segments (2.3%), including 5 RCA segments, 2 LAD segments, and 6 LCx segments.
Of 575 segments in group 2, images without motion artifacts (score 1) were obtained in 217 segments (37.7%); images with minor artifacts (score 2), in 237 segments (41.2%); and images with moderate artifacts (score 3), in 85 segments (14.8%). Severe artifacts (score 4) occurred in 36 segments (6.3%), including 15 RCA segments, 6 LAD segments, and 15 LCx segments.
Interobserver agreement for image-quality rating was good (k = 0.72). Table III summarizes the image-quality scores for all of the segments and coronary arteries.
There were no statistically significant differences in diagnostic-image quality score (scores 1–3) between group 1 and group 2, although scores were higher in group 2 (1.53 ± 0.43 vs 1.58 ± 0.33; P = 0.561). There were also no statistically significant differences between the groups in the image quality of the RCA (P = 0.415), the LCA (P = 0.966), or the LCx (P = 0.091), although better results in group 2 were observed for the LCA than for the RCA or LCx. Obvious degradation in the visual imaging quality of all coronary arteries was observed in 4 patients in group 2 whose heart rates exceeded 100 beats/min (102–108 beats/min): the image-quality score for any coronary segment in these 4 patients was no better than 2, even for the left main trunk.
In the 27 patients who underwent invasive coronary angiography, 392 segments were evaluated for stenosis (5 segments were missing because of anatomic variations, and 8 segments had a diameter of less than 1.5 mm at their origins). The DSCT identified 117 significant stenoses (29.8%): 52 in group 1 and 65 in group 2. In comparison with conventional coronary angiography, the diagnostic accuracy of the DSCT was 96.2% (152 of 158 segments) in group 1 and 94.9% (222 of 234 segments) in group 2. There was no statistically significant difference in diagnostic accuracy between the groups. Of 18 segments that produced false-positive and false-negative findings in both groups, 13 (72.2%) were affected by severe calcification (Fig. 2).
In group 1, no significant correlation was found between variability of heart rate and the diagnostic image quality of all coronary segments (r = 0.184, P = 0.267). In group 2, however, a moderate correlation was found, albeit not a strong one (r = 0.494, P = 0.001) (Fig. 3). In group 2, no significant correlation was present between heart-rate variability and diagnostic-image quality score for the RCA (r = 0.238, P = 0.109) or for the LCA (r = 0.175, P = 0.241); however, there was a significant albeit weak correlation for the LCx (r = 0.362, P = 0.016). Fourteen high-heart-rate patients with 36 nonevaluable segments had a mean heart rate of 86.5 ± 7.2 beats/min (range, 73–106 beats/min) and a mean heart-rate variability of 5.1 ± 2.2 beats/min (range, 5.6–9.7 beats/min). Eight low-heart-rate patients with 13 nonevaluable segments displayed heart-rate variability from 5.9 to 8.7 beats/min. In both groups, all of the segments that could not be evaluated were found in patients whose heart rates varied more than 5 beats/min. All of these segments were distal segments and branches of the RCA, or its middle segment.
The positions of the mean optimal systolic and diastolic reconstruction windows in relation to heart rate are shown in Figure 4. At heart rates of or slower than 70 beats/min, the evaluative image quality occurred best in diastole (70%–80% of the R–R interval) in 394 of the 545 coronary segments (72.3%). At heart rates faster than 70 beats/min, the best reconstruction time shifted to end-systole (35%–45% of the R–R interval) at progressively higher heart rates: the best evaluative image quality occurred in systole in 486 of the 539 coronary segments (90.2%). Of the 40 patients whose heart rates exceeded 70 beats/min, 3 patients whose heart rates were between 70 and 75 beats/min (mean respective heart rates, 71, 72, and 74 beats/min) displayed the optimal reconstruction interval at diastole. Regarding the selection of the optimal reconstruction interval for the coronary segments, interobserver agreement was good (k = 0.76).
When 64-MDCT has been used, studies have shown that CT coronary angiography can be performed with diagnostic image quality at a wide range of heart rates.6 However, almost all studies upon 64-section CT have revealed a persistent, significant impact of heart rate on image quality,3,18–20 and the image quality of the LCx has reflected a weak dependence on the average heart rate.3,21 In contrast, our results upon DSCT without the patients' use of β-blocker premedication showed that high heart rates have no effect on image quality. By including patients who had heart rates as fast as 108 beats/min, we found minor-to-moderate blurring of vascular walls due to motion artifacts that were present at high heart rates. We found no significant image-quality degradation in any coronary segment or in any individual coronary artery in our high- or low-heart-rate patients. By selecting the optimal reconstruction phases for each vascular segment, we found diagnostic image quality in 539 of the 575 coronary segments in our group 2 (93.7%). Our data support previous reports13,17 that the diagnostic image quality of DSCT is high in patients whose heart rates range from 71 to 99 beats/min.
Early feasibility studies22,23 showed DSCT coronary angiography to be highly accurate in the diagnosis of coronary artery disease in patients who have high heart rates, in comparison with conventional coronary angiography. Although our sample was small, our results also indicated that DSCT preserves high diagnostic accuracy in such patients. Regarding the detection of significant coronary stenoses, we determined diagnostic accuracies of 96.2% in 152 of 158 arterial segments in group 2, and 94.9% in 222 of 234 segments in group 1. According to our results, 73% of false-positive and false-negative findings were due to severe calcification in the arterial segments that were studied. Hence, calcification was the chief inhibitor of diagnostic accuracy. This finding is in agreement with previous data upon 64-section CT8 and DSCT.22
In 3 studies,3,24,25 significant negative correlations between heart-rate variability and image quality for the entire coronary arterial tree and for each coronary artery were observed upon 64-detector row CT. These results contrast with the findings discussed in a DSCT study by Matt and colleagues.17 By using a monosegment reconstruction algorithm, the investigators found no significant effect of inter-heartbeat variation on overall image quality. According to our Pearson correlation coefficient analysis, however, a significant effect of inter-heartbeat variation on the image quality of all segments can be observed in patients who have high heart rates. This difference may be because our study focused on high-heart-rate patients; Matt and colleagues studied patients who had low and high heart rates. We also observed that variability of heart rate had a significant effect on the image quality of the LCx in patients who had high heart rates. Our finding differs from the usual conclusion that the RCA has the least stationary time (especially when heart rate increases), for this reason: at higher heart rates, diastole shortens more than systole,25 and the image quality of the LCx (whose best reconstruction interval is mid-diastole) is lowered accordingly. This becomes more notable when high heart rates are combined with heart-rate variability during scanning. However, the effect on the image quality of the RCA and the LCA is not obvious, because imaging of the RCA in late systole and early diastole is less prone to motion artifacts when shortening of diastole occurs. In both of our groups, all 22 patients with all 49 nonevaluable segments exhibited heart-rate variability of more than 5 beats/min (range, 5.6–9.7 beats/min). Variability of more than 5 beats/min contributed most strongly to the inability to evaluate arterial segments.
At higher heart rates, the motion-free time in middiastole shortens more than that in mid-to-late systole.24 The shortening of diastole in patients with high heart rates explains a correlation between the motion artifacts and heart rate.2 Upon the use of 64-MDCT, middiastolic reconstruction was preferred when heart rates were 75 or even 85 beats/min.3,20,25,26 This increase in heart-rate threshold may be explained by the increase in temporal resolution: when DSCT of the temporal resolution remains at 83 ms, the transition of the optimal reconstruction interval shifts from diastole to systole. Matt and colleagues found that the best reconstruction intervals in diastole and systole, respectively, were 70% and 30% for heart rates of 50 to 60 beats/min, 75% and 35% for rates of 61 to 80 beats/min, and 85% and 40% for rates faster than 80 beats/min. They also found that the transition of the optimal reconstruction interval from diastole to systole occurred at approximately 80 beats/min.17 In our DSCT study, the findings were little different. We observed that the positions of the optimal reconstruction windows were 70% to 80% diastolic and 30% to 40% systolic in our group 1, and 70% to 80% diastolic and 35% to 45% systolic in our group 2. We also observed that all but 3 of the optimal reconstruction intervals of patients whose heart rates exceeded 70 beats/min occurred in systole. Even in 15 of our 40 patients whose heart rates were ≤70 beats/min, the optimal reconstruction interval occurred in systole. These findings suggest that the transition point from diastolic to systolic reconstructions should be >70 beats/min.
Our study has the following limitations. First, the image-quality scoring system may have been influenced by a subjectivity bias. Second, the small number of patients whose heart rates exceeded 100 beats/min may have limited our descriptive statistics.
Address for reprints: Yi Huan, MD, Department of Radiology, Xijing Hospital, Fourth Military Medical University, 17# West Changle Road, Xi'an 710032, Shaanxi Province, PRC. E-mail: ten.haey@0003iynauh