Jacobo Kirsch Ivan Buitrago Tan-Lucien H. Mohammed Tianming Gao Craig R. Asher Gian M. Novaro

DOI 10.1007/s10554-011-9928-9 ORIGINAL PAPER Detection of coronary calcium during standard chest computed tomography correlates with multi-detector computed tomography coronary artery calcium score Jacobo Kirsch Ivan Buitrago Tan-Lucien H. Mohammed Tianming Gao Craig R. Asher Gian M. Novaro Received: 12 March 2011 / Accepted: 15 July 2011 Ó Springer Science+Business Media, B.V. 2011 Abstract The correlation between formal coronary artery calcium scoring (CACS) determined by multidetector CT (MDCT) and the presence of coronary calcium on standard non-gated CT chest examinations was evaluated. In 163 consecutive healthy participants, we performed screening same-day standard non-gated, non-enhanced CT chest exams followed by high-resolution, ECG-synchronized MDCT exams for CACS. For the standard CT examinations, a scoring system (Weston score, range 0 12) was developed assigning a score (0 3) for each coronary vessel including the left main trunk. Overall, 30% and 39% of patients had CAC on standard CT and MDCT exams, respectively (P = 0.13). CAC on J. Kirsch (&) Division of Radiology, Cleveland Clinic Florida, 2950 Cleveland Clinic Boulevard, Weston, FL 33331, USA e-mail: kirschj@ccf.org I. Buitrago Department of Internal Medicine, Cleveland Clinic Florida, Weston, FL, USA T.-L. H. Mohammed Imaging Institute, Cleveland Clinic, Cleveland, OH, USA T. Gao Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, OH, USA C. R. Asher G. M. Novaro Department of Cardiology, Cleveland Clinic Florida, Weston, FL, USA standard CT was highly correlated to the Agatston CACS on the MDCT (Spearman correlation coefficient 0.83, P \ 0.001). Absence of calcium on the standard CT exam was associated with a very low CACS (mean Agatston 0.5, range 0 19). A Weston score[2 identified acacs[ 100 with an area under the curve of 0.976, sensitivity of 100%, and specificity of 85%. A Weston score[7 identified a CACS [ 400 with an area under the curve of 0.991, sensitivity of 100%, specificity of 98%. The intra-observer variability was low as was the interobserver variability between a cardiac specialized radiologist and a non-specialized reader. A visual coronary artery scoring system on standard, non-gated CT correlates well with traditional methods for CACS. Further, a non-expert cardiac radiologist performed equally well to a cardiac expert. This information suggests that a visual scoring system, at least in a descriptive manner can be utilized for a general statement about coronary artery calcification seen on standard CT imaging to guide clinicians in risk stratification. Keywords Coronary artery disease Computed tomography Risk assessment Coronary calcification Introduction Risk stratification is the cornerstone of preventive cardiology. Traditionally, a clinical risk model such

as the Framingham risk score has been utilized to stratify risk (low, intermediate or high) and thus determine the aggressiveness of management [1]. Various laboratory tests and imaging modalities have been studied in attempt to improve standard risk estimation. Coronary artery calcium (CAC) scoring (CACS) has been demonstrated, with the support of large prospective studies to add incremental prognostic information to predict cardiovascular events relative to clinical risk models [2 6]. The American College of Cardiology/American Heart Association consensus statement regarding CACS supports the use of CACS in selected, asymptomatic, intermediate-risk patients [7 9]. However, widespread utilization of CACS has not occurred due to concerns related to radiation exposure, cost and the potential for unjustified further testing [10, 11]. Furthermore, there is little evidence to validate outcomes of asymptomatic patients treated for abnormal CACS. As a result of these unresolved issues, most health insurance plans in the United States do not cover CACS and therefore testing is not commonly done. With a rapidly growing utilization of standard chest computed tomography (CT) for innumerable reasons, radiologists have recognized that calcium can be readily seen in most vascular structures including the coronary arteries. The significance of coronary calcium seen during these examinations is not well-established. Therefore, we sought to evaluate the correlation between CACS determined by multi-detector computed tomography (MDCT) and the presence of calcium on standard non-enhanced, non-gated CT examinations of the chest using a visual scoring method (Weston score). Secondly, we assessed the accuracy of a cardiac trained radiologist (expert reader) and a non-cardiac radiologist for determining CACS. Patients and methods Patients In accordance with the policies set by our internal institutional review board, we performed a retrospective review of the charts of patients consecutively referred for total body CT examinations from July 2008 to December 2009 as part of the Executive Health Program at our institution. Patients with prior thoracic surgery (cardiac or non-cardiac) and/or with pacemaker or defibrillator wires were excluded from the study to avoid obscuration of the coronary arteries from the associated streak artifact. Information on patient demographics and clinical conditions was collected, and is shown in Table 1. CT examinations Patients underwent screening same-day standard nongated, non-enhanced CT examinations of the chest immediately followed by high-resolution, ECG-synchronized MDCT examinations for CACS. All studies were conducted on either a 16-slice or 64-slice MDCT system (Siemens Healthcare, Forchheim, Germany). The technical parameters of the acquisition for both the standard non-gated, non-enhanced CT of the chest and the native sequence for CACS are outlined in Table 2. All reconstructions were transferred to a PC-based workstation (Syngo CaScoring, Wizard; Siemens Healthcare, Forchheim, Germany) for quantification of coronary calcifications. The values of the marginal and diagonal branches were excluded in the analysis. Coronary artery calcifications were defined as more than 2 adjacent pixels with absorption values of more than 130 HU, and quantitative CAC was calculated Table 1 Demographic and clinical characteristics based on Weston score Characteristic Total cohort (n = 163) CAC present (n = 49) CAC absent (n = 114) P Value Age (years) 51 ± 9 57 ± 8 48 ± 8 \0.001 Men 127 (78%) 42 (86%) 85 (75%) 0.12 Body mass index (kg/m 2 ) 27 ± 4 28 ± 4 27 ± 4 0.76 Hypertension 21 (13%) 8 (16%) 13 (11%) 0.39 Hypercholesterolemia 91 (56%) 34 (69%) 57 (50%) 0.03 Diabetes mellitus 3 (2%) 2 (4%) 1 (1%) 0.22 Current smoker 44 (27%) 11 (23%) 33 (29%) 0.39

Table 2 CT acquisition parameters Standard chest CT Coronary calcification 16-row 64-row 16-row 64-row Scan type Spiral Spiral Spiral Spiral Rotation time (ms) 0.5 0.33 0.5 0.33 Collimation 16 9 0.75 64 9 0.6 16 9 1.5 24 9 1.2 kvp 120 120 120 120 Reference mas 180 180 200 250 CARE dose 4D ON ON OFF OFF Recon kernel B40f B40f B35 B35 Slice thickness (mm) 5 5 3 3 Recon increments (mm) 5 5 3 3 FOV (mm) 300 300 250 250 by the Agatston method [12]. A CT technologist with cardiac imaging training (6 years experience) performed all studies. The standard CT examinations were analyzed visually using mediastinum soft tissue window settings (WW 400, WL 40). A scoring system (Weston score) was developed assigning a score for each major coronary vessel (the left main trunk, left anterior descending artery, left circumflex artery, right coronary artery), as follows: 0 no visually detected calcium; 1 if only a single high density pixel was detected; 3 if the calcium was dense enough to create blooming artifact; and 2 for calcium in between 1 and 3 (Fig. 1). The Weston score was calculated by the sum of the score for each vessel (range 0 12). In addition to the evaluation of all the cases by observer 1 (4 years of cardiac imaging experience), a random sample of 95 CT scans were reviewed independently by a second observer (9 years of thoracic imaging experience). For measurement of intra-observer variability, observer 1 performed a second reading of 93 cases 1 month apart. The order of the cases reviewed differed between the 2 sessions. During the study, both observers were blinded to the results of the calcium score examinations and the patients demographic and clinical data. Statistical analysis Baseline patient characteristics were compared between CAC present and CAC absent group based on the standard CT examinations. For continuous outcomes, student t test was used. For categorical outcomes, Chi-square or Fisher s exact test was used. Results of CT exams were also compared between CAC present and absent group; Fisher s exact test or Chi-square test was used to calculate the P-values. The Agatston calcium score and the Weston score for each coronary vessel and their sum were compared. Correlation of the scores between the two types of exams was measured using the Pearson correlation test and the nonparametric Spearman s correlation test. A P value of 0.05 or less was considered to indicate a statistically significant difference. The inter-observer and intra-observer reliability for Weston score measurements was assessed using the intraclass correlation coefficient and concordance correlation coefficient, considering values [0.75 to represent excellent agreement. Receiver operating characteristic (ROC) curve analysis was performed for total CACS at cutoffs of 100 and 400. The optimal cutoff point of the Weston score to predict a calcium score greater than 100 and 400 was identified from the ROC analysis. Sensitivity, specificity, positive predictive value, and negative predictive value were chosen at the optimum cutoff point. Additionally, baseline patient demographics and clinical data shown in Table 1 were added to the ROC prediction model, and tested if such addition improved the discrimination. All statistics analysis was conducted in SAS 9.2 (Cary, NC). The study was approved by the Institutional Review Board at Cleveland Clinic Florida, Weston, FL.

Fig. 1 Axial standard CT of the chest images (a, c, and e) paired with their corresponding images from the ECG-gated CT of the heart (b, d, and f, respectively). Images a and b are from a 62-year-old male with a total calcium score of 5.1 and show a punctate fossae of increased attenuation at the mid right coronary artery graded as a Weston score of 1. Images c and d are from a 52-yearold male with a total calcium score of 262.2 and show scattered (nonblooming) calcified plaque involving the proximal left anterior descending artery graded as a Weston score of 2. Images e and f are from a 57-year-old male with a total calcium score of 94.5 and show foci of blooming calcified plaque at the mid left anterior descending artery graded as a Weston score of 3 Results There were 127 men (mean ± SD age, 51 ± 9 years; range, 25 75 years) and 36 women (51 ± 9 years; 25 67 years) in the study. The baseline patient characteristics are shown in Table 1. All patients were asymptomatic as stipulated in the Executive Health Program protocol; 65% of patients had at

Fig. 2 Total coronary artery calcium by Agatston score plotted against the Weston score (Pearson s correlation coefficient 0.81, P \ 0.001; nonparametric Spearman s coefficient 0.83, P \ 0.001) least one cardiovascular risk factor other than sex and age. Total Agatston CACS was plotted against the Weston score (Fig. 2). The Pearson s correlation coefficient between total CACS and the Weston score was 0.81 (P \ 0.001) whereas the nonparametric Spearman s coefficient was 0.83 (P \ 0.001). The distribution of total CACS as assessed by the Weston score at each level is shown in Table 3. The distribution of patients by total Agatston CACS is as follows: 1 100 (n = 46), 101 399 (n = 13),[400 (n = 4). Overall, 30% (n = 49/163) and 39% (n = 63/163) of patients had CAC on standard CT and MDCT exams, respectively (P = 0.13). Absence of calcium on the standard CT exam was associated with a very low CACS (mean Agatston score 0.5, range 0 19). In general, mean total calcium score increased with increasing Weston score. Coronary calcium scores were compared per-vessel between MDCT and the Weston score. Using the Spearman correlation coefficient, the correlations with the Weston score were as follows: left main trunk = 0.35 (P \ 0.001); left anterior descending = 0.82 (P \ 0.001); left circumflex = 0.55 (P \ 0.001); and right coronary artery = 0.62 (P \ 0.001). Figure 3 shows the ROC curve for patients with a total calcium score [100. The area under the curve (the concordance index) was 0.9760. The best cutoff point to predict a calcium score [100 or not was when the predicted probability of a total score [100 was 0.07, which corresponded to a Weston score of 2. This cutoff point predicted a CACS [ 100 with 100% sensitivity (17/17), 85% specificity (124/146), 44% positive predictive value (17/39), and 100% negative predictive value (124/124). Mean CACS of patients with a Weston score \ versus C 2 was 1.0 ± 4.0 versus 158.3 ± 237.6 (P \ 0.001). Figure 4 shows the ROC curve for patients with a total calcium score [400. The area under the curve (the concordance index) was 0.9914. The best cutoff point to predict a calcium score [400 or not was Table 3 Summary of total coronary artery calcium scores by each Weston score Weston score N Coronary artery calcium scores Mean ± SD Minimum Maximum 0 114 0.5 ± 2.3 0 18.8 1 10 6.7 ± 10.6 0 33.3 2 16 46.6 ± 39.3 11.2 135.1 3 5 77.9 ± 77.2 5.1 198.2 4 5 84.4 ± 45.9 3.6 117.4 5 4 138 ± 99.6 34.1 262.2 6 2 82.4 ± 25.8 64.1 100.6 7 1 413.1 413.1 413.1 8 3 447.5 ± 296.5 198.1 775.4 10 2 487.3 ± 230.5 324.3 650.3 11 1 1,170.8 1,170.8 1,170.8 Fig. 3 Receiver operating characteristic curve analysis reveals a cutoff Weston score of 2 to predict a total CACS [ 100. The area under the curve (AUC) was 0.9760

Fig. 4 Receiver operating characteristic curve analysis reveals a cutoff Weston score of 7 to predict a total CACS [ 400. The area under the curve (AUC) was 0.991 when the predicted probability of a total score [400 was 0.17, which corresponded to Weston score of 7. This cutoff point predicted a CACS [ 400 with 100% sensitivity (4/4), 98% specificity (156/159), 57% positive predictive value (4/7), and 100% negative predictive value (156/156). Mean CACS of patients with a Weston score \ versus C 7 was 15.4 ± 39.5 versus 557.3 ± 334.7 (P \ 0.001). For both ROC analyses, the baseline characteristics were added as additional predictors to the model. Then, the area under the ROC curves was compared before and after the added predictors; none of the baseline variables significantly improved the area under the ROC curve. Two observers rated 95 samples, with each of them observing Weston scores on all 95 samples. The inter-observer reliability was very good with an interobserver correlation coefficient of 0.932. Also one observer rated 93 samples at two time points on each sample. The intra-observer reliability was very good with an intra-observer correlation coefficient of 0.963 and a concordance correlation coefficient of 0.962. Discussion The primary finding of this study is that a visual coronary artery scoring system (Weston score) on standard, non-gated CT correlates well with traditional methods for CACS. This correlation remains strong for total and per-vessel scoring. A Weston score of 2 was associated with a CACS [ 100 and a Weston score of 7 associated with a CACS [ 400. The addition of baseline clinical variables did not improve the discrimination of the Weston score to predict the level of CACS. Furthermore, per vessel correlation of CAC was good, although weakest for the left main trunk. A secondary finding of this study is that the intra-observer reliability was high as was the inter-observer reliability between a cardiac specialized radiologist and a non-specialized reader. The Framingham risk model remains the standard tool for predicting the likelihood of a cardiac event in asymptomatic individuals. It allows clinicians to determine the need for medical therapies (e.g. aspirin, statin) and whether further testing is justified (e.g. stress testing). However, risk prediction with the Framingham or other clinical scores remains inaccurate in many patients and efforts are ongoing to improve estimation of events. CACS has emerged as an additive prognostic indicator of myocardial infarction and cardiac death in intermediate risk patients. Its wide-spread utilization has been tempered by concerns over safety related to radiation exposure, cost and the possibility of further overutilization of testing. Furthermore, there are still no prospective studies to prove that the detection and management of subclinical disease affects outcomes. Nonetheless, the American College of Cardiology/American Heart Association writing group in 2007 considered CACS in intermediate-risk patients to be reasonable since it may reclassify patients into a high risk category and modify treatment [9]. Standard chest CT is increasingly utilized for numerous reasons, particularly in the emergency room setting. With the growing recognition of the clinical importance of coronary and vascular calcification, radiologists are more commonly reporting on the presence of any such calcification. Concomitantly, several investigations have sought to determine the significance of CAC on standard, non-gated CT of the chest. Wu et al. showed that low-dose ungated MDCT for lung cancer screening was reliable for determining the presence of CAC. In their study of 483 subjects, risk stratification by Agatston scoring was highly concordant with low-dose un-gated MDCT and dedicated regular-dose ECG-gated

MDCT. However, in their study they analyzed both the gated and un-gated studies using calcium scoring dedicated software [13]. To our knowledge, only 3 other studies have attempted to visually quantify the amount of CAC on un-gated CT exams of the chest. In 2 studies by Shemesh et al., the authors showed that CACS can be generated from an un-gated low-dose CT scan, and may provide prognostic information in regards to cardiovascular death. They developed an ordinal scoring system based on the length of the artery involved as absent, mild, moderate, and severe. In a small subgroup of their population (16 patients), they also found an excellent correlation (r = 0.84, P \ 0.0001) between their ordinal scores and the Agatston scores [14]. The visually assessed CACS was found to be a significant predictor of cardiovascular death, independent of age, gender, and history of tobacco use [15]. A study recently published by Einstein et al. showed that CACS can be visually assessed from low-dose CT scans (used for attenuation correction during cardiac single-photon emission computed tomography/ct myocardial perfusion imaging) with high agreement with standard Agatston scores. In this study, experienced readers assigned visually the degree of coronary calcification using a scale of scores. The study also showed a high degree of inter-observer reproducibility of the visually obtained scores, with readers reporting identical scores in 65% of cases and scores within 1 category of each other in 93% of cases [16]. However, it should be noted that in this study and the other similar studies, the readers were dedicated experienced cardiac imagers, whereas in our study one of the observers was not a dedicated cardiac radiologist. Study limitations The present study has several limitations worth discussing. First, our population was predominantly a low-risk cohort, and therefore the spectrum of patients with higher CAC scores is under-represented. A second potential limitation involves imaging processing. The post-processing algorithm used to create the standard CT images of the chest may vary between scanners and/or institutions as some centers are using thinner than 5 mm slice thickness and sharper kernel algorithms which may influence the visual assessment proposed in this study. The per vessel correlation of CACS was good though weakest for the left main trunk. This may be attributable to difficulties discerning the length and bifurcation of the left main trunk on non-contrast images such that calcification may have been scored in the left anterior descending or circumflex vessel. In conclusion, our data support the feasibility of CACS during standard CT imaging of the chest. With the increasing utilization of chest CT examinations, we believe that a visual assessment of CAC at least in a descriptive fashion should become a customary addition to the interpretation of standard CT scans. Although the impact of early detection of subclinical CAC remains uncertain, the clinician is provided with additive information regarding risk stratification that may aid in determining aggressiveness of care. Conflict of interest References None. 1. Wilson PW, D Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB (1998) Prediction of coronary heart disease using risk factor categories. Circulation 97:1837 1847 2. Budoff MJ, Gul KM (2008) Expert review on coronary calcium. Vasc Health Risk Manag 4:315 324 3. Raggi P, Gongora MC, Gopal A, Callister TQ, Budoff M, Shaw LJ (2008) Coronary artery calcium to predict allcause mortality in elderly men and women. J Am Coll Cardiol 52:17 23 4. Budoff MJ, Shaw LJ, Liu ST, Weinstein SR, Mosler TP, Tseng PH, Flores FR, Callister TQ, Raggi P, Berman DS (2007) Long-term prognosis associated with coronary calcification: observations from a registry of 25, 253 patients. J Am Coll Cardiol 49:1860 1870 5. Greenland P, LaBree L, Azen SP, Doherty TM, Detrano RC (2004) Coronary artery calcium score combined with Framingham score for risk prediction in asymptomatic individuals. JAMA 291:210 215 6. Detrano R, Guerci AD, Carr JJ, Bild DE, Burke G, Folsom AR, Liu K, Shea S, Szklo M, Bluemke DA, O Leary DH, Tracy R, Watson K, Wong ND, Kronmal RA (2008) Coronary calcium as a predictor of coronary events in four racial or ethnic groups. N Engl J Med 358:1336 1345 7. Johnson KM, Dowe DA (2010) The detection of any coronary calcium outperforms Framingham risk score as a first step in screening for coronary atherosclerosis. AJR Am J Roentgenol 194:5 1243 8. Elias-Smale SE, Proença RV, Koller MT, Kavousi M, van Rooij FJ, Hunink MG, Steyerberg EW, Hofman A, Oudkerk M, Witteman JC (2010) Coronary calcium score

improves classification of coronary heart disease risk in the elderly: the Rotterdam study. J Am Coll Cardiol 56: 1407 1414 9. Greenland P, Bonow RO, Brundage BH, Budoff MJ, Eisenberg MJ, Grundy SM, Lauer MS, Post WS, Raggi P, Redberg RF, Rodgers GP, Shaw LJ, Taylor AJ, Weintraub WS; American College of Cardiology Foundation Clinical Expert Consensus Task Force (ACCF/AHA Writing Committee to Update the 2000 Expert Consensus Document on Electron Beam Computed Tomography); Society of Atherosclerosis Imaging and Prevention; Society of Cardiovascular Computed Tomography (2007) ACCF/ AHA 2007 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 2000 expert consensus document on electron beam computed tomography) developed in collaboration with the society of atherosclerosis imaging and prevention and the society of cardiovascular computed tomography. J Am Coll Cardiol 49:378 402 10. Kim KP, Einstein AJ, Berrington de Gonzalez A (2009) Coronary artery calcification screening: estimated radiation dose and cancer risk. Arch Intern Med 169:1188 1194 11. Shaw LJ, Min JK, Budoff M, Gransar H, Rozanski A, Hayes SW, Friedman JD, Miranda R, Wong ND, Berman DS (2009) Induced cardiovascular procedural costs and resource consumption patterns after coronary artery calcium screening: results from the EISNER (early identification of subclinical atherosclerosis by noninvasive imaging research) study. J Am Coll Cardiol 54:1258 1267 12. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R (1990) Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 15:827 832 13. Wu MT, Yang P, Huang YL, Chen JS, Chuo CC, Yeh C, Chang RS (2008) Coronary arterial calcification on lowdose ungated MDCT for lung cancer screening: concordance study with dedicated cardiac CT. AJR Am J Roentgenol 190:923 928 14. Shemesh J, Henschke CI, Farooqi A, Yip R, Yankelevitz DF, Shaham D, Miettinen OS (2006) Frequency of coronary artery calcification on low-dose computed tomography screening for lung cancer. Clin Imaging 30:181 185 15. Shemesh J, Henschke CI, Shaham D, Yip R, Farooqi AO, Cham MD, McCauley DI, Chen M, Smith JP, Libby DM, Pasmantier MW, Yankelevitz DF (2010) Ordinal scoring of coronary artery calcifications on low-dose CT scans of the chest is predictive of death from cardiovascular disease. Radiology 257:541 548 16. Einstein AJ, Johnson LL, Bokhari S, Son J, Thompson RC, Bateman TM, Hayes SW, Berman DS (2010) Agreement of visual estimation of coronary artery calcium from lowdose CT attenuation correction scans in hybrid PET/CT and SPECT/CT with standard Agatston score. J Am Coll Cardiol 56:1914 1921