Assessment of Agatston Coronary Artery Calcium Score Using Contrast-Enhanced CT Coronary Angiography

Transcription

1 Cardiopulmonary Imaging Original Research van der Bijl et al. Agatston and Coronary Angiography Cardiopulmonary Imaging Original Research Noortje van der Bijl 1 Raoul M. S. Joemai 1 Jacob Geleijns 1 Jeroen J. Bax 2 Joanne D. Schuijf 2 Albert de Roos 1 Lucia J. M. Kroft 1 van der Bijl N, Joemai RMS, Geleijns J, et al. Keywords: Agatston score, coronary artery calcium, angiography, unenhanced DOI:1.2214/AJR Received October 2, 29; accepted after revision January 3, Department of Radiology, Leiden University Medical Center, C2-S, Albinusdreef 2, 2333 ZA Leiden, The Netherlands. Address correspondence to L. J. M. Kroft (L.J.M.Kroft@lumc.nl). 2 Department of Cardiology, Leiden University Medical Center, The Netherlands. AJR 21; 195: X/1/ American Roentgen Ray Society Assessment of Agatston Coronary Artery Calcium Using Contrast-Enhanced Coronary Angiography OBJEIVE. The purpose of this article is to evaluate to what extent Agatston scores may be derived from coronary angiography (A) examinations, compared with traditional unenhanced calcium scores. MATERIALS AND METHODS. Fifty patients with a calcium score Agatston score of zero and 5 patients with a calcium score Agatston score of 1 or greater whose calcium scores had been calculated and who had undergone A using volumetric 32- MD were included. Agatston scores were obtained at 3.-mm slices for calcium score and A. Method agreement, interobserver agreement, and diagnostic performance of A for detecting coronary calcium were evaluated. RESULTS. Of 5 patients with a positive calcium score Agatston score, coronary artery calcium was detected with A in 43 patients by observer 1 (mean A score, 12 ± 22; mean calcium score, 254 ± 51) and in 46 patients by observer 2 (mean A score, 94 ± 147; mean calcium score, 272 ± 531). Of the 5 patients with a calcium score Agatston score of zero, 49 (98%, observer 1) and 5 (1%, observer 2) had a zero score with A as well. An intraclass correlation of.78 and.62 was found between calcium score and A (p <.1), whereas higher Agatston scores were underestimated with A. For observer 1, the sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy for detection of coronary calcium with A were 86%, 98%, 98%, 88%, and 92%, respectively, and the corresponding values for observer 2 were 92%, 1%, 1%, 93%, and 96%, respectively. Interobserver agreement was.996 for calcium score and.93 for A. CONCLUSION. Coronary artery calcium can be detected on A images with high accuracy. The Agatston calcium score derived from A images shows good correlation with unenhanced calcium score and is highly reproducible. However, higher Agatston scores are systematically underestimated when derived from A images. C oronary artery disease is one of the leading causes of death. Quantifying the amount of coronary artery calcium with unenhanced calcium score has been shown to be a reliable noninvasive technique for screening risk of future cardiac events [1, 2] and can be quantified by using the Agatston score [3] or scores such as the volume score [4] or calcium mass [5]. Large patient studies have shown that the amount of coronary artery calcium based on the Agatston score is a strong predictor for risk of myocardial infarction and sudden cardiac death, independently of conventional coronary risk factors [6 8]. Absent or low coronary artery calcium has been shown to be highly accurate in exclud- ing coronary artery disease in asymptomatic patients [9, 1]. However, the value of a zero or low calcium score in symptomatic patients remains less clear. Several studies have reported the presence of obstructive ( 5%) noncalcified plaque in up to 8.7% of symptomatic patients with zero or low calcium scores [11 14]. Therefore, in symptomatic patients, calcium score may be followed by angiography (A), or A may be performed alone. A with the aid of IV contrast agent injection is widely used for the evaluation of suspected coronary artery disease. A confirms or excludes significant coronary artery stenosis with high accuracy compared with invasive coronary angiography [15 17]. Furthermore, it has been shown that the presence AJR:195, December

2 van der Bijl et al. of coronary artery disease detected with A is of incremental and independent value in predicting all-cause mortality in symptomatic patients [18 2]. A allows visualization of the vessel lumen but also of the vessel wall, including calcified atherosclerotic plaque [21]. However, the level of contrast enhancement in the coronary vessels may obscure plaque and may obviate reliable measurements of plaque density, especially in noncalcified plaques. It is conceivable that the amount of coronary calcium may be estimated by using the A images owing to the relatively high density of calcified plaques compared with that of noncalcified plaques. Only a few studies have addressed this issue previously [22 24]. To derive calcium scores from A, these studies increased the threshold values for coronary calcium from the standard (i.e., 13 HU) to 35 HU [22, 23] or even 6 HU [24], to avoid luminal contrast being falsely depicted as coronary artery calcium. The increase in attenuation threshold resulted in underestimation [22, 24] or overestimation of the calcium scores [23], whereas increasing threshold values may lead to decreased sensitivity for depicting small amounts of coronary calcium [25]. The purpose of this study was to evaluate to what extent Agatston scores may be derived from A examinations compared with traditional calcium score. Materials and Methods Study Population Fifty patients with an Agatston score of zero and 5 patients with a positive Agatston score ( 1) were, per group, consecutively selected from a database of patients who had undergone both unenhanced calcium score and contrast-enhanced A investigations for clinical indications (59 men and 41 women; mean age, 55 ± 11 years; height, 175 ± 2 cm; weight, 82 ± 15 kg). Patients with coronary stents (n = 5), pacemakers (n = 6), and prosthetic heart valves (n = 1) had been excluded beforehand to avoid scoring artifacts. Our institutional review board does not require its approval for anonymous retrospective technical analysis of data, as was the case in this study. Protocol All examinations were performed with a 32- MD scanner (Aquilion ONE, Toshiba). The patients had undergone prospectively ECG-gated unenhanced volumetric (to calculate calcium score) for scoring the amount of coronary calcium according to Agatston et al. [3] during the same session, followed by a prospectively ECGgated contrast-enhanced volumetric A for coronary artery evaluation, with (n = 68) or without (n = 32) functional analysis. To lower the heart rate, 25 1 mg of oral metoprolol was administered to patients with a cardiac frequency exceeding 6 beats per minute, if no contraindications were present. Mean (± SD) heart rate during scanning was 54 ± 7 beats per minute. The scan range was planned between the carina and cardiac apex. Depending on the expected scan range, a 32.5 mm or 28.5 mm detector configuration was used. Immediately before image acquisition, an optimal reconstruction phase was determined during a breath-hold exercise with ECG recording. Full cardiac calcium score acquisition was performed within a single heart beat during breath-hold at inspiration. Scan parameters were 12 kv tube voltage and 2 4 ma tube current (mean, 32 ± 49 ma), depending on patient size and shape (2 ma for small or thin patients, 25 ma for average size patients, and 3 4 ma for large or obese patients). Rotation time was.35 second. Effective radiation dose estimation was based on the dose length product provided by the scanner for each patient and by using the correction factor.17 for chest imaging in adults [26]. The estimated dose was 1.9 ±.3 msv. A was performed after bolus injection of 5 7 ml of iodinated contrast agent (4 mg/ml iomeprol; Iomeron, Bracco) via antecubital vein injection with a flow rate of 5. ml/s followed by a 2 ml mix of 5% contrast agent and 5% saline, followed by a 25-mL saline flush using an automatic injector (Stellant, MedRad). Bolus tracking was performed by placing a region of interest in the left ventricle. Image acquisition was automatically started 7 seconds after reaching a predefined threshold difference of 1 HU. Scan parameters dependent on body mass index (BMI) were as follows: 1 kv and ma for BMI (15 patients); 12 kv and 4 58 ma for BMI 23 3 (65 patients); and 135 kv and 51 ma for BMI greater than 3 (2 patients). Mean BMI was 26 ± 4. Rotation time was.35 second. The scan range for A had been planned with the aid of the calcium score scan; care was taken to include the full range of the coronary arteries. Full cardiac A acquisition was performed within a single heartbeat during breath-hold at inspiration, with or without functional analysis, including dose modulation throughout the cardiac cycle. Estimated effective radiation dose was 1.7 ± 5.9 msv. Image Reconstruction Standard reconstruction kernel filters were used for image reconstruction: FC12 for calcium score and FC43 for A. For calcium score, nonoverlapping 3.-mm data sets were reconstructed, which is the standard method used in clinical practice based on electron-beam [3]. Similarly, for A, data sets of nonoverlapping 3.-mm slices were reconstructed for evaluation of coronary calcium. An additional.5-mm A data set, with.25-mm increments, that is used for A evaluation in clinical practice was reconstructed. The reconstructions were transferred to a workstation for analysis. Analysis Analysis of coronary artery calcium was performed on a postprocessing workstation (Vitrea FX, version 1., Vital Images) using dedicated calcium score analysis software (V, Vital Images). Coronary calcium was defined as an area of at least three face-connected voxels in the axial plane in the course of a coronary artery, with an attenuation threshold value of 13 HU or greater. Three in-axial-plane face-connected voxels correspond to a minimum lesion area greater than 1 mm 2, which is used as a reference value in calcium scoring [6]. Calcium scores of each investigation were calculated and expressed as Agatston scores for standard of reference calcium score and for A reconstructions. Contour drawing was performed by two investigators with 2 years (observer 1) and 4 years (observer 2) of experience in cardiac ; observer 1 was supervised by a radiologist with 7 years of experience in cardiac. Observers were aware that patients had been selected on the basis of the presence (n = 5) or absence (n = 5) of coronary calcium but were not aware of medical history of individual patients. Examinations were presented in random order. For A, calcifications that were visually identified in the course of the coronary arteries were marked by the investigator. The investigator was allowed to compare the 3.-mm A data set with the.5-mm A data set. Marking was done on the 3.-mm data set by precise contour drawing of visually identified calcium spots after zooming in on the focus, allowing visual identification of individual pixels. Automatic recognition of pixels of 13 HU or greater was switched off because the design of the coronary calcium analysis program is dedicated for unenhanced. After manual contour drawing, calcium scores were automatically calculated on the basis of the 13 HU threshold value. After obtaining the calcium scores in the A data sets of all patients, the calcium scores were obtained in calcium score data sets. With automatic recognition of 13 HU or more, pixels exceeding this threshold value are colored purple by the postprocessing tool. In each slice (depending on scan range, 47 or 53 slices), these areas were manually 13 AJR:195, December 21

3 Agatston and Coronary Angiography TABLE 1: Calcium and Agatston Risk Group Distribution Risk Group, Agatston No. of Patients Calcium encircled when present in the course of each coronary artery. Calcium scores were automatically calculated. A time interval of at least 2 weeks between scoring the A and the calcium score was used to prevent recognition bias. Patients were classified according to Agatston risk groups as defined by Rumberger et al. [27]. Although this risk stratification scheme is based on absolute Agatston scores and does not account for patient age, sex, and race, recent studies have shown that absolute calcium scores may predict cardiovascular events better than adjusted percentiles [28, 29]. Observer 1 Observer 2 Angiography p Calcium Angiography 5 ±.3 ± 1.6 NS ±.1 ±.8 NS ± 3 4 ± 7 NS 3 ± 4 4 ± 4 NS ± ± 2 NS 45 ± ± ± ± 53 < ± 1 1 ± 56 <.1 > ± ± ± ± Note Data are the mean ± SD Agatston score for 3. mm calcium score and 3. mm angiography reconstructions. NS = not significant. Statistical Analysis Statistical analysis was performed using SPSS for Windows (version 16., SPSS). The diagnostic performance of A in the detection of coronary artery calcium is presented as sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy; calcium score was used as the standard of reference. The mean and median calcium scores and SDs were calculated for calcium score and for A for the whole group and for the Agatston risk groups [27]. The Wilcoxon s signed rank test for two related samples was p Fig year-old man with coronary artery calcifications. A and B, calcium score images show coronary artery calcifications (arrows) in left anterior descending coronary artery. calcium score was calculated with automatic recognition of 13 HU switched off (A) and automatic recognition switched on (purple color, B). Agatston calcium score calculated by calcium score was 151. C and D, Coronary artery calcifications (arrows) in left anterior descending coronary artery are seen on coronary angiography (A) images with 3. mm (C) and.5 mm (D) reconstruction. Agatston calcium score was 96 with 3. mm A reconstruction (C). Comparison with.5 mm A reconstruction that is used in clinical practice for coronary artery lumen evaluation was allowed (D). applied to determine a statistically significant difference between the calcium scores obtained with calcium score and A. The intraclass correlation coefficient (ICC) was calculated to evaluate method agreement between calcium score and A investigations and to assess interobserver agreement. An ICC less than.4 indicated poor reproducibility, an ICC of.4.75 indicated fair to good reproducibility, and an ICC greater than.75 indicated excellent reproducibility [3]. The method described by Bland and Altman [31] was used to study limits of agreement and systematic error between the two methods and between the two observers, respectively. A p value of less than.5 was considered statistically significant. Results Positive Calcium ( 1) For observer 1, the mean calcium score of the 5 patients with a positive calcium score was 254 ± 51 (median, 82). At A, coronary calcium was detected in 43 (86%) of 5 patients, with a mean calcium score of 12 ± 22 (median, 4). Seven patients had a false-negative calcium score at A, with scores of 1 (n = 4 patients), 2 (n = 1 patient), 4 (n = 1 patient), and 8 (n = 1 patient). Figure 1 shows a comparison of coronary artery calcium visualization with calcium score and A. For observer 2, the mean calcium score of the 5 patients with a positive calcium score was 272 ± 531 (median, 82). Coronary artery calcium was detected in 46 (92%) of these 5 patients at A, with a mean calcium score of 94 ± 147 (median, 45). Four patients had a false-negative calcium score, with scores of 1 (n = 2 patients) and 8 (n = 2 patients). The distribution of patients within different Agatston risk groups, as defined by Rumberger et al. [27], is shown in Table 1. No statistically significant difference was found between calcium score and A-derived calcium scores for patients with low calcium scores AJR:195, December

4 van der Bijl et al. Difference Observer 1 (CS A) A 2, 1, 1, Mean Agatston Observer 1 ([CS + A] /2) (< 1, both observers). For the patients with calcium scores exceeding 1 (observer 2) and 1 (observer 1), the calcium score was statistically significantly underestimated by A, with a mean factor of 2.8 (observer 1 range, ; observer 2 range, ) compared with the reference method (Table 1). Despite underestimated calcium scores by A, good ICCs (observer 1,.78; observer 2,.62) were found between calcium score and A-derived calcium scores, showing good-to-excellent agreement between the two methods for the whole patient group (both p <.1). Bland-Altman plots in Figure 2 show the limits of agreement and systematic error for individual scores and show good agreement for low calcium scores, whereas the highest scores (exceeding 466 for observer 1 and 596 for observer 2; for the highest risk Agatston group, > 4) show systematic Difference Observer 2 (CS A) 2, 1, 1, , 1, , 1, error resulting from underestimation of Aderived calcium scores. Zero Calcium () For observer 1, 49 (98%) of the 5 patients with a zero calcium score had a zero score derived from A as well. In one patient, a calcium score of 5 was calculated. In that patient, contrast enhancement in the left anterior descending artery was visually interpreted as coronary calcium. All 5 patients with a zero calcium score scored by observer 2 had a zero A-derived calcium score as well. Mean Agatston Observer 2 ([CS + A] /2) Fig. 2 Bland-Altman analysis showing limits of agreement and systematic errors for both observers. Results for Agatston calcium scores obtained with calcium score (CS) and angiography (A) are shown for observer 1 (A) and for observer 2 (B). Dashed lines show upper and lower limits of agreement ± 1.96 SD and 95% CI. Note that good agreement can be observed for low calcium scores, whereas highest Agatston scores above upper limit of calcium score exceeding 544 (observer 1) and 686 (observer 2) contain systematic error. TABLE 2: Shift in Risk Group Distribution for Calcium and Angiography Derived Agatston s Risk Group, Agatston No. of Patients Observer 1 Observer 2 Shift Up Shift Down Total Shift Shift Up Shift Down Total Shift > Total Note Data the number of patients shifting between Agatston score risk groups. Dashes indicate that no risk group-shift occurred. B Diagnostic Performance for Detecting Coronary Artery Calcium For observer 1, the diagnostic performance and predictive value of A for the detection of coronary artery calcium were as follows: sensitivity, 86% (95% CI, 73 94%); specificity, 98% (95% CI, 88 1%); positive predictive value, 98% (95% CI, 86 1%); negative predictive value, 88% (95% CI, 75 94%); and diagnostic accuracy, 92%. For observer 2, the values were as follows: sensitivity, 92% (95% CI, 8 97%); specificity, 1% (95% CI, 91 1%); positive predictive value, 1% (95% CI, 9 1%); negative predictive value, 93% (95% CI, 81 98%); and diagnostic accuracy, 96%. The change in classification according to risk groups for the whole patient population for A-derived Agatston scores, compared with the standard of reference calcium score, for the two observers is shown in Table 2. Classification in another risk group occurred for 27% of patients by using A-derived calcium scores; for observer 1, downgrading of the Agatston risk group occurred in 23% of cases, and upgrading of Agatston risk group occurred in 4% of cases. For observer 2, classification in another risk group occurred for 2% of patients by using A-derived calcium scores, whereas downgrading of the Agatston risk group occurred in 19% of cases, and upgrading of Agatston risk group occurred in 1% of cases. Interobserver Variability Interobserver agreement was excellent for both the calcium score Agatston scores (ICC,.997; p <.1) as well as the Aderived Agatston scores (ICC,.94; p <.1) for the total group. Also, for the groups with a positive calcium score Agatston score, 132 AJR:195, December 21

5 Agatston and Coronary Angiography 5 5 Difference CS Agatston (Observer 1 Observer 2) A , Mean CS Agatston ([Observer 1 + Observer 2]/2) excellent interobserver agreement was found (ICC for calcium score,.996; ICC for A,.93; both p <.1). Bland-Altman plots in Figure 3 show levels of agreement and systematic error for individual scores and show excellent observer agreement. In the highest risk group, a few outliers for calcium score Agatston scores were found between the two observers. The differences in Agatston scores between the two observers for these outliers can be explained by dissimilar inclusion of coronary artery calcium at the origin of the coronary arteries that was continuous with calcifications in the aortic wall. Discussion The main finding of the current study is that the presence of coronary calcification can be estimated by using contrast-enhanced A, with excellent diagnostic accuracy, positive predictive value, and specificity. Second, the Agatston score derived from coronary A correlates well with unenhanced calcium score, which is the standard of reference. Third, both enhanced and unenhanced provide equivalent Agatston scores when there is a limited amount of coronary calcium, but coronary A underestimates the amount of calcium in cases with higher Agatston scores. Fourth, deriving Agatston scores from coronary A can be obtained with a high reproducibility and excellent observer agreement. The excellent diagnostic accuracy, positive predictive value, and specificity found in the current study indicate that coronary calcium Difference A Agatston (Observer 1 Observer 2) 25 can be considered as present with a positive A-derived Agatston score. Although a good sensitivity and negative predictive value were found, the absence of coronary calcium on A may be a false-negative observation of coronary calcium that is actually present. However, no large amounts of coronary calcium were missed on A, because in missed cases, the actual Agatston score was low, and within Agatston risk group 1 or less, the median Agatston score was 2. In two previous studies in which Agatston scores were derived from A investigations, the detection threshold for coronary calcium was increased from 13 to 35 HU for Aderived calcium scores and was compared with the 13 HU calcium score threshold [22, 23]. One study, which used nonoverlapping 3.-mm-slice calcium score to compare with overlapping 1.25-mm-slice A in 5 patients, reported an overestimation of A Agatston scores by a factor of approximately 3 [23]. It is unclear whether the overestimation found in that study was due to the inclusion of contrast material exceeding 35 HU or to differences in reconstructed slice thickness and use of overlapping image reconstruction. With thin-slice overlapping images, voxel sizes and, thus, partial volume effect decrease. The chance of a voxel containing sufficient calcification attenuation to reach the detection threshold increases with smaller voxels [32], leading to higher scoring results [25, 33, 34]. Another study used 3.-mm slices and 2.-mm increments for both A and , Mean A Agatston ([Observer 1 + Observer 2]/2) Fig. 3 Bland-Altman analysis showing interobserver agreement, limits of agreement, and systematic errors for Agatston calcium score calculated by both calcium score (CS) (A) and angiography (A) (B). Dashed lines show upper and lower limits of agreement ± 1.96 SD and 95% CI. Excellent agreement can be observed for both CS as well as A-derived calcium scores between two observers (p <.1). B calcium score [22]. In that study, A data for only 28 of 38 patients were used for analysis, because seven patients with a negative calcium score were not included for A analysis, and A analysis could not be performed for another three patients. A Agatston scores were underestimated by a factor of approximately 3 [22], which is the range of the current study. In the current study, the traditional threshold value of 13 HU was used for both A-derived calcium scores and calcium score. Good agreement was found between calcium score and Aderived calcium scores. Although low Agatston scores derived from A were not statistically different from the calcium score, a substantial underestimation was found in the higher risk groups, which led to a downshift of risk group for 2 22% of the patients, whereby 1 4% of patients shifted from the high-risk group (calcium score, > 4) to the intermediate-risk group (calcium score, 1 4). Total shift in risk groups occurred for 2 27% of all our patients by visual assessment of coronary artery calcium and by using the 13 HU threshold value, compared with 57% of analyzed patients in a study that used automatic assessment and 35 HU as threshold value for detection [23]. Furthermore, in the current study, all 1 A studies were included for analysis. Moreover, by using volumetric imaging within a single heartbeat for both the calcium score and A acquisitions, the 3D volume data sets reconstructed to 3.-mm slices were technically most comparable. It AJR:195, December

6 van der Bijl et al. should be noted that the original Agatston score and large outcome studies were based on nonoverlapping 3.-mm-slice data sets as well [3, 1]. In symptomatic patients, the extent of coronary artery calcium has been shown to provide additional prognostic information over invasive coronary angiography alone [6]. However, even a zero calcium score may not exclude obstructive coronary artery disease [11 14], and a positive score is no direct indicator for coronary artery stenosis. Therefore, calcium score alone seems not to be optimal for excluding coronary artery disease in symptomatic patients and is often followed by A. Now that the A technique has developed into a clinical tool that is increasingly used for coronary artery evaluation in routine clinical practice, A rather than calcium score may be used for coronary artery evaluation. A allows direct evaluation of the presence and extent of coronary artery luminal obstruction, whereas Abased estimation of the presence and extent of coronary artery calcium from the same images may provide additional risk information that may obviate the traditional calcium score. Radiation exposure is of major concern in coronary A because of the associated risk of radiation-induced cancer [35]. Several methods, including prospective ECG-triggering techniques, have been developed and have been very effective in reducing patient dose [36, 37]. If a separate calcium score examination can be avoided by using the A examination in deriving the presence and extent of coronary calcium, this may aid substantially in further decreasing patient dose. The present study has some limitations. Deriving Agatston scores from A was more time consuming than deriving Agatston scores from calcium scores (the standard of reference), with approximately double the time needed for analysis, because the reader was not alerted by automatic color encoding of coronary artery calcium during evaluation. Also, the scoring method used may not be available on all workstations and may be vendor dependent. Application of the technique in routine clinical practice may require software improvements. Further coronary artery analysis software developments may facilitate deriving calcium scores from A investigations. Furthermore, although good method agreement between calcium score and A-derived calcium scores was found, it is unclear whether the A-derived Agatston scores may be used for actual risk stratification even if a conversion factor is applied as is done now for the large patient databases obtained by electron beam [9, 1]. A relatively small group of 1 patients was used for analysis in this feasibility study, and patients were selected according to the presence or absence of coronary calcium (n = 5 patients each). It is not known what the effect of using A-derived calcium scores by means of risk stratification and clinical consequences would be for individual patients, because such studies would require large study populations other than our selected groups of 5 patients each, because outcome depends on disease prevalence. It should be noted that patient risk stratification is not based on the amount of coronary artery calcium alone but also depends on patient characteristics and the presence or absence of other risk factors [2]. It has been shown that using A data, with its information on luminal narrowing and plaque composition, has incremental prognostic value over using calcium score alone [14, 18, 38]. Therefore, combining the results of luminal narrowing, plaque composition, and calcium score may provide optimal information for -based risk stratification. Radiation dose was relatively high, because the study was performed early after installation of the new scanner type. At that time, the majority of patients (68%) had undergone imaging that included a functional analysis with a relatively high radiation dose compared with prospective A for coronary imaging alone that is routinely applied today. Also, tube current settings for calcium score have now been substantially decreased. In conclusion, coronary artery calcium can be detected on A with high accuracy. The Agatston calcium score derived from A shows good correlation with unenhanced calcium score and is highly reproducible. However, higher Agatston scores are systematically underestimated when derived from A. Acknowledgment We thank B. J. A. Mertens for statistical advice. References 1. Oudkerk M, Stillman AE, Halliburton SS, et al. Coronary artery calcium screening: current status and recommendations from the European Society of Cardiac Radiology and North American Society for Cardiovascular Imaging. Int J Cardiovasc Imaging 28; 24: Greenland P, Bonow RO, Brundage BH, et al. ACCF/AHA 27 clinical expert consensus document on coronary artery calcium scoring by computed tomography in global cardiovascular risk assessment and in evaluation of patients with chest pain: a report of the American College of Cardiology Foundation Clinical Expert Consensus Task Force (ACCF/AHA Writing Committee to Update the 2 Expert Consensus Document on Electron Beam Computed Tomography). Circulation 27; 115: Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 199; 15: Callister TQ, Cooil B, Raya SP, Lippolis NJ, Russo DJ, Raggi P. Coronary artery disease: improved reproducibility of calcium scoring with an electron-beam volumetric method. Radiology 1998; 28: Hoffmann U, Siebert U, Bull-Stewart A, et al. Evidence for lower variability of coronary artery calcium mineral mass measurements by multidetector computed tomography in a communitybased cohort consequences for progression studies. Eur J Radiol 26; 57: Keelan PC, Bielak LF, Ashai K, et al. Long-term prognostic value of coronary calcification detected by electron-beam computed tomography in patients undergoing coronary angiography. Circulation 21; 14: Wong ND, Hsu JC, Detrano RC, Diamond G, Eisenberg H, Gardin JM. Coronary artery calcium evaluation by electron beam computed tomography and its relation to new cardiovascular events. Am J Cardiol 2; 86: Arad Y, Spadaro LA, Goodman K, Newstein D, Guerci AD. Prediction of coronary events with electron beam computed tomography. J Am Coll Cardiol 2; 36: Shaw LJ, Raggi P, Schisterman E, Berman DS, Callister TQ. Prognostic value of cardiac risk factors and coronary artery calcium screening for all-cause mortality. Radiology 23; 228: Budoff MJ, Shaw LJ, Liu ST, et al. Long-term prognosis associated with coronary calcification: observations from a registry of 25,253 patients. J Am Coll Cardiol 27; 49: Cheng VY, Lepor NE, Madyoon H, Eshaghian S, Naraghi AL, Shah PK. Presence and severity of noncalcified coronary plaque on 64-slice computed tomographic coronary angiography in patients with zero and low coronary artery calcium. Am J Cardiol 27; 99: Akram K, O Donnell RE, King S, Superko HR, 134 AJR:195, December 21

7 Agatston and Coronary Angiography Agatston A, Voros S. Influence of symptomatic status on the prevalence of obstructive coronary artery disease in patients with zero calcium score. Atherosclerosis 29; 23: Rubinshtein R, Gaspar T, Halon DA, Goldstein J, Peled N, Lewis BS. Prevalence and extent of obstructive coronary artery disease in patients with zero or low calcium score undergoing 64-slice cardiac multidetector computed tomography for evaluation of a chest pain syndrome. Am J Cardiol 27; 99: van Werkhoven JM, Schuijf JD, Gaemperli O, et al. Incremental prognostic value of multi-slice computed tomography coronary angiography over coronary artery calcium scoring in patients with suspected coronary artery disease. Eur Heart J 29; 3: Mowatt G, Cook JA, Hillis GS, et al. 64-Slice computed tomography angiography in the diagnosis and assessment of coronary artery disease: systematic review and meta-analysis. Heart 28; 94: Miller JM, Rochitte CE, Dewey M, et al. Diagnostic performance of coronary angiography by 64-row. N Engl J Med 28; 359: Meijer AB, YL O, Geleijns J, Kroft LJ. Metaanalysis of 4- and 64-MD angiography for assessing coronary artery stenosis. AJR 28; 191: Ostrom MP, Gopal A, Ahmadi N, et al. Mortality incidence and the severity of coronary atherosclerosis assessed by computed tomography angiography. J Am Coll Cardiol 28; 52: Min JK, Shaw LJ, Devereux RB, et al. Prognostic value of multidetector coronary computed tomographic angiography for prediction of all-cause mortality. J Am Coll Cardiol 27; 5: Hadamitzky M, Freissmuth B, Meyer T, et al. Prognostic value of coronary computed tomographic angiography for prediction of cardiac events in patients with suspected coronary artery disease. JACC Cardiovasc Imaging 29; 2: Budoff MJ, Achenbach S, Blumenthal RS, et al. Assessment of coronary artery disease by cardiac computed tomography: a scientific statement from the American Heart Association Committee on Cardiovascular Imaging and Intervention, Council on Cardiovascular Radiology and Intervention, and Committee on Cardiac Imaging, Council on Clinical Cardiology. Circulation 26; 114: Muhlenbruch G, Wildberger JE, Koos R, et al. Coronary calcium scoring using 16-row multislice computed tomography: nonenhanced versus contrast-enhanced studies in vitro and in vivo. Invest Radiol 25; 4: Hong C, Becker CR, Schoepf UJ, Ohnesorge B, Bruening R, Reiser MF. Coronary artery calcium: absolute quantification in nonenhanced and contrast-enhanced multi-detector row studies. Radiology 22; 223: Glodny B, Helmel B, Trieb T, et al. A method for calcium quantification by means of coronary angiography using 64-multidetector : very high correlation with Agatston and volume scores. Eur Radiol 29; 19: Muhlenbruch G, Klotz E, Wildberger JE, et al. The accuracy of 1- and 3-mm slices in coronary calcium scoring using multi-slice in vitro and in vivo. Eur Radiol 27; 17: Menzel HG, Schibilla H, Teunen D, et al. European guidelines on quality criteria for computed tomography: report no. EUR EN. Luxembourg: European Commission, Rumberger JA, Brundage BH, Rader DJ, Kondos G. Electron beam computed tomographic coronary calcium scanning: a review and guidelines for use in asymptomatic persons. Mayo Clin Proc 1999; 74: Budoff MJ, Nasir K, McClelland RL, et al. Coronary calcium predicts events better with absolute calcium scores than age-sex-race/ethnicity percentiles: MESA (Multi-Ethnic Study of Atherosclerosis). J Am Coll Cardiol 29; 53: Akram K, Voros S. Absolute coronary artery calcium scores are superior to MESA percentile rank in predicting obstructive coronary artery disease. Int J Cardiovasc Imaging 28; 24: Rosner B. Multisample inference. Fundamentals of biostatistics, 5th ed. Belmont: Duxbury Press, 2: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: Muhlenbruch G, Thomas C, Wildberger JE, et al. Effect of varying slice thickness on coronary calcium scoring with multislice computed tomography in vitro and in vivo. Invest Radiol 25; 4: Sabour S, Rutten A, van der Schouw YT, et al. Inter-scan reproducibility of coronary calcium measurement using multi detector-row computed tomography (MD). Eur J Epidemiol 27; 22: Groen JM, Greuter MJ, Schmidt B, Suess C, Vliegenthart R, Oudkerk M. The influence of heart rate, slice thickness, and calcification density on calcium scores using 64-slice multidetector computed tomography: a systematic phantom study. Invest Radiol 27; 42: Einstein AJ, Henzlova MJ, Rajagopalan S. Estimating risk of cancer associated with radiation exposure from 64-slice computed tomography coronary angiography. JAMA 27; 298: Earls JP, Berman EL, Urban BA, et al. Prospectively gated transverse coronary angiography versus retrospectively gated helical technique: improved image quality and reduced radiation dose. Radiology 28; 246: Efstathopoulos EP, Kelekis NL, Pantos I, et al. Reduction of the estimated radiation dose and associated patient risk with prospective ECG-gated 256-slice coronary angiography. Phys Med Biol 29; 54: Rubinshtein R, Halon DA, Gaspar T, Peled N, Lewis BS. Cardiac computed tomographic angiography for risk stratification and prediction of late cardiovascular outcome events in patients with a chest pain syndrome. Int J Cardiol 29; 137: AJR:195, December