Author + information
- Received October 3, 2008
- Revision received November 6, 2008
- Accepted November 12, 2008
- Published online April 1, 2009.
- Martin Hadamitzky, MD⁎,⁎ (, )
- Barbara Freißmuth, MD⁎,
- Tanja Meyer, MD⁎,
- Franziska Hein, MD⁎,
- Adnan Kastrati, MD⁎,
- Stefan Martinoff, MD†,
- Albert Schömig, MD⁎ and
- Jörg Hausleiter, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. med. Martin Hadamitzky, Deutsches Herzzentrum, Lazarettstrasse 36, München 80636, Germany
Objectives We assessed the rate of cardiac events after detection or exclusion of obstructive coronary artery disease (CAD) by coronary computed tomography angiography (CCTA).
Background Several studies have demonstrated a high diagnostic accuracy of CCTA for detection of obstructive CAD compared with invasive angiography, but data regarding the clinical prognostic value of CCTA are limited.
Methods In all, 1,256 consecutive patients with suspected CAD undergoing 64-slice CCTA in our institution between October 2004 and September 2006 were observed prospectively for the occurrence of severe cardiac events (cardiac death, myocardial infarction, or unstable angina requiring hospitalization: primary study end point) and all cardiac events (additionally including revascularization >90 days after CCTA). The observed rate of all cardiac events was compared with the event rate predicted by the Framingham risk score. Obstructive CAD was defined as ≥50% diameter stenosis in any coronary artery.
Results During a median follow-up of 18 months (interquartile range 14 to 25 months), the overall rates of severe and all cardiac events were 0.6% and 1.8%, respectively. In 802 patients without obstructive CAD, there were 4 cardiac events, of which 1 was severe, whereas in 348 patients with obstructive CAD, there were 17 cardiac events, of which 5 were severe. The difference between the 2 groups was highly significant both for severe events (odds ratio: 17.3, 95% confidence interval: 3.6 to 82.5) and for all cardiac events (odds ratio: 16.1, 95% confidence interval: 7.2 to 36.0; both p < 0.001). The rate of all cardiac events in patients without obstructive CAD was significantly lower than predicted by the Framingham risk score (p = 0.01).
Conclusions In patients with suspected CAD, CCTA has a significant prognostic impact on the prediction of cardiac events for the subsequent 18 months. The exclusion of obstructive CAD by CCTA identifies a patient population with an event risk lower than predicted by conventional risk factors.
Cardiovascular disease is the most common cause of morbidity and mortality in Western countries (1). The availability of effective treatment options both by medical therapy and interventional as well as surgical procedures makes it important to identify patients at risk and equally important to identify patients not needing treatment. Noninvasive imaging techniques play an important role as a gatekeeper for invasive diagnostic and therapeutic procedures. A good prognostic accuracy and an excellent negative predictive value for cardiac events have particularly been demonstrated for stress perfusion imaging techniques such as nuclear myocardial perfusion imaging and magnetic resonance stress perfusion imaging (2,3).
During the last few years coronary computed tomography angiography (CCTA) has emerged as a new technique for noninvasive 3-dimensional visualization of the coronary arteries. Although several studies showed good correlation between CCTA and invasive coronary angiography (4), concerns remain regarding its prognostic value. The number of studies investigating this topic is very limited (5–9). Therefore, the aim of this prospective observational study was to determine the predictive value of CCTA for cardiovascular events in patients with suspected coronary artery disease (CAD) and to compare the prognostic value of CCTA with the pre-test risk assessed by the Framingham risk score.
We prospectively enrolled all consecutive patients undergoing CCTA on a 64-slice CT scanner (Siemens Somatom Sensation 64 Cardiac, Siemens, Forchheim, Germany) in our institution from October 1, 2004, to September 15, 2006. Written informed consent was obtained from all patients before the investigation. Patients were eligible for this study if CAD was not previously known, but suspected. Exclusion criteria comprised patients investigated in an acute life-threatening condition and patients without stable sinus rhythm. A structured interview was performed before the investigation, and information about age, height, and weight of the patient, cardiac history, and current medication was collected. The following cardiac risk factors were recorded: 1) presence and degree of hypertension (for binary analysis, a systolic blood pressure >140 mm Hg was considered as abnormal regardless of antihypertensive therapy); 2) diabetes mellitus (defined as fastening blood glucose level >7 mmol/l or use of oral antidiabetic therapy or insulin); 3) smoking (defined as current smoker or previous smoker within the last year); and 4) a positive family history (defined as presence of CAD in first-degree relatives <55 years of age if male and <65 years of age if female). In addition, laboratory tests for total cholesterol, low-density lipoprotein and high-density lipoprotein fractions, and triglycerides were performed. Finally, 2 prognostic scores were calculated: the Morise pre-test score (10) and the Framingham risk score with the established categorical model using low-density lipoprotein cholesterol according to Wilson et al. (11). The study design was approved by the local ethics committee.
The detailed CT scan protocol was described elsewhere (12). In brief, to lower the heart rate at the time of the CT study, up to 4 doses of 5 mg metoprolol were administered intravenously to patients with a heart rate >60 beats/min. All patients with a systolic blood pressure of at least 100 mm Hg received nitroglycerin 0.8 mg sublingually for coronary vasodilatation. Images for calcium scoring were acquired by a noncontrast-enhanced scan and analyzed with a commercially available software package (Siemens CalciumScore, Siemens, Erlangen, Germany).
Contrast timing was tested by an initial bolus-timing scan using 20 cc of contrast (Iomeprol, Imeron 350, Bracco Altana Pharma GmbH, Konstanz, Germany), iodine content 350 mg/cc, followed by a 50-cc saline chaser. The contrast-enhanced scan was obtained using 80 to 140 cc of contrast individually adapted to the selected table feed and scan range at a rate of 4 to 5 cc/s followed by a 50-cc saline chaser. The scan was performed using a 32 × 0.6 mm detector with oscillating beam, leading to an effective resolution of 64 × 0.4 mm (13). For reduction of radiation dose exposure, an electrocardiographically gated tube current modulation was used in patients with stable sinus rhythm. Images were reconstructed in mid-diastole with individually optimized position of the reconstruction window. Additional image reconstructions were performed in end-systole if required. A dataset of axial slices, multiplanar reformations, and 3 perpendicular sets of thin-slab maximum intensity projections orientated along the heart axis (5-mm thickness, 1-mm increment) were reconstructed and investigated for the presence of obstructive CAD.
The coronary artery tree was segmented according to the modified American Heart Association classification (14). Each segment with a diameter >1.5 mm was evaluated visually by 2 experienced readers (1 radiologist and 1 cardiologist) for the degree of lumen narrowings, and rated semiquantitatively by 4 groups: <25%, 25% to 49%, 50% to 74%, and ≥75%. Differences in the evaluation between the 2 readers were resolved by common consensus. If severe calcification or other artifacts made a valid assessment impossible, the segment was classified as not evaluable. The result of the CCTA was defined as nonobstructive if all evaluated segments had no lumen narrowing ≥50% and no artifacts leading to nondiagnostic image quality. The CCTA was defined as obstructive if at least 1 segment had a lumen narrowing ≥50% or if a relevant lumen narrowing could not be ruled out because of severe artifacts. All patients with obstructive CAD by CCTA were advised to undergo invasive angiography.
Follow-up information was obtained either by clinical visits, telephone contact, or by detailed questionnaires sent by mail. All reported events were verified by hospital records or direct contacts with the attending physician.
The following clinical events were recorded: 1) cardiac death (including death without definitive cause); 2) nonfatal myocardial infarction; 3) unstable angina pectoris requiring hospitalization; and 4) coronary revascularization (either by bypass surgery or percutaneous coronary intervention). In addition, information on noncardiac deaths was collected. Coronary revascularizations occurring early after CCTA were obviously performed as a result of CCTA because all patients with obstructive CAD by CCTA were recommended to have invasive angiography and intervention as needed. Therefore, coronary revascularization procedures performed earlier than 90 days after CCTA were excluded from further analysis. For follow-up analysis, all patients undergoing early revascularization were censored at the time of the intervention and therefore excluded from further analysis.
The primary end point of this study was the composite end point of “severe cardiac events” including cardiac death, nonfatal myocardial infarction, and unstable angina requiring hospitalization. The secondary end point was the combined end points of “all cardiac events,” additionally including revascularizations later than 90 days after CCTA, representing the events defining CAD in the Framingham risk score (11).
The prognostic value of a noninvasive test like CCTA for CAD is strongly dependent on the pre-test risk. To assess the additional value of the CCTA both for detecting and excluding obstructive CAD, we compared the observed rate of any cardiac events with the event rate predicted by the Framingham risk score, one of the best-evaluated scores for assessing the risk for developing CAD (15,16). Because the Framingham risk score denotes the risk for developing symptomatic CAD within 10 years expressed in percentages, the individual risk for a cardiac event within the follow-up period was calculated as Framingham risk score/1,000 multiplied by the individual follow-up period in years. For the patient group without obstructive CAD and the group with obstructive CAD, the cumulative predicted event rates were then calculated as the sum of the individual risks and compared with the observed event rate.
Categorical variables were expressed as frequencies and percentages; comparison between 2 groups was done by Fisher's exact test. Continuous variables were expressed as median and interquartile range (first and third quartile) and compared by Wilcoxon rank sum test. The occurrence of combined end points was analyzed by the Kaplan-Meier method, and odds ratios were calculated with the log-rank test; for continuous variables, they were calculated between values above and below median. Predictive values were denoted with their 95% confidence interval (CI) calculated according to the modified Wald method (17). Statistical significance was accepted for p values <0.05. The software package R (version 2.6.1) was used for statistical analysis (18).
Study population and follow-up
In 1,667 patients, CCTA was performed during the study period. Of these, 411 had known CAD: 166 patients had a history of myocardial infarction, 217 patients had a previous bypass surgery, 15 patients had a coronary stent implanted, and 13 patients had otherwise known CAD. In addition, 5 patients had CCTA in acute life-threatening conditions (4 aortic dissections and 1 infective endocarditis), and 56 patients had no stable sinus rhythm during the investigation. These 411 patients were excluded by protocol. From among the remaining 1,195 patients, a clinical follow-up could be established for 1,150 (96.2%). The median duration of follow-up was 540 days, with an interquartile range from 411 to 762 days and a total range from 303 to 1,333 days. All evaluations are based on this patient group (Fig. 1).
The median patient age was 60.2 years, with an interquartile range of 52.1 to 66.1 years; 795 of the 1,150 (69%) patients were male. The patient characteristics and coronary risk factor profile mirrors a study population of low to intermediate risk for CAD. The detailed patient characteristics and indications for CCTA are summarized in Table 1. Patients with arrhythmias had either exercise-induced premature ventricular beats during stress testing or short runs of nonsustained ventricular tachycardias during Holter electrocardiographic monitoring or intermittent atrial arrhythmias with increasing frequency within the last year. “Other indications” mainly consisted of exclusion of CAD before noncardiac surgery. All patients with coronary plaques or stenoses, irrespective of the extent, were advised to minimize their cardiovascular risk profile. Whenever indicated, we advised patients to stop smoking and to vigorously treat diabetes, and recommended acetylsalicylate, statins, and antihypertensive medication.
Invasive angiography and early revascularizations
As a result of CCTA, 348 (30%) patients were advised to undergo invasive coronary angiography, 129 (11%) because of inconclusive findings and 219 (19%) because of obstructive CAD. Of these 348 patients, 215 (62%) underwent invasive angiography within 90 days after CCTA, which was followed by revascularization procedures in 119 (34%) patients, 6 (2%) by coronary artery bypass graft and 113 (32%) by percutaneous coronary intervention. All patients undergoing early revascularization were censored at the time of the intervention for the event-free survival analysis. In 95 (27%) patients, obstructive CAD was ruled out by invasive angiography, and 2 (1%) with obstructive CAD were treated medically; 132 (38%) patients did not undergo invasive angiography.
Severe cardiac events
A total of 6 severe cardiac events occurred during follow-up, which compares with an overall event rate of 0.4% during the first year after CCTA. The severe cardiac events are summarized in Table 2. Significantly more severe cardiac events were observed among patients with obstructive CAD (5 vs. 1 severe cardiac events, respectively, in patients with obstructive vs. nonobstructive CAD; p < 0.001 in the log-rank test) (Fig. 2), resulting in an odds ratio for severe cardiac events of 17.3 (95% CI: 3.6 to 82.5). For comparison, the odds ratios for the presence of typical angina pectoris, the Morise pre-test score, and the Framingham risk score were 4.5 (95% CI: 0.6 to 32.2), 1.7 (95% CI: 0.4 to 7.1), and 8.6 (95% CI: 2.1 to 35.3), respectively; for the calcium score, the odds ratio was 5.1 (95% CI: 0.7 to 34.9). The event rate in the first year after CCTA calculates to 0.1% (95% CI: 0.05 to 1.2) for patients with nonobstructive CAD and to 1.7% (95% CI: 0.6 to 4.5) for patients with obstructive CAD.
All cardiac events
During follow-up, significantly more late (>90 days after CCTA) coronary revascularization procedures were performed in patients with obstructive CAD (15 vs. 3 revascularization procedures in patients with obstructive vs. no obstructive CAD, respectively; p < 0.001) (Table 2). With an overall event rate during the first year after CCTA of 1.1% for all cardiac events, this difference translates to an odds ratio of 16.1 (95% CI: 7.2 to 36.0) (Fig. 3). For comparison, the odds ratios for the presence of typical angina pectoris, the Morise pre-test score, and the Framingham risk score were 2.0 (95% CI: 0.5 to 8.3), 2.6 (95% CI: 1.1 to 6.1), and 3.3 (95% CI: 1.5 to 7.6), respectively; for the calcium score, the odds ratio was 7.7 (95% CI: 2.7 to 21.8). The event rate during the first year after CCTA calculated to 0.3% (95% CI: 0.1 to 0.8) in patients without obstructive CAD and to 4.3% (95% CI: 2.3 to 7.9) in patients with obstructive CAD.
Looking at the stenosis grading, these events were distributed as follows: 2 events among 498 patients with no stenosis ≥25%, 2 events among 304 patients with mild stenosis (25% to 49%), 11 events in 159 patients with a moderate stenosis (50% to 74%), 3 events among 64 patients with high-grade stenoses (≥75%), and 3 events among 126 patients with nonevaluable segments. The pre-defined cutoff of 50% lumen narrowing proved to be most effective for event prediction as compared with levels of 25% and 75% of lumen narrowing. Although all 3 levels had a significant prognostic value, the odds ratio of the 50% cutoff of 16.1 was higher than those of the cutoffs of 25% and 75% (9.4 and 6.5, respectively).
Comparison with conventional cardiovascular risk scoring
The predicted risk according to the Framingham risk score for patients without obstructive CAD calculated to 12 events, with a 95% CI ranging from 6 to 19 events. In this patient group, 4 events were observed. Accordingly, significantly fewer events than predicted by the Framingham risk score occurred in this group (p = 0.01).
In patients with obstructive CAD, the number of predicted events was 8, with a 95% CI ranging from 3 to 14 events. Here, the number of 17 observed events was significantly higher than predicted (p = 0.003) (Fig. 4).
One patient without obstructive CAD by CCTA died of pneumonia; in the patient group with obstructive CAD, 3 patients died of malignancies, 2 of septicemia, and 1 of an embolic cerebral insult.
The key findings of this study are: 1) that patients with obstructive CAD by CCTA suffer significantly more frequently from severe as well as from all cardiac events than do patients without obstructive CAD; and 2) that CCTA that excludes obstructive CAD identifies a population with a very low risk for severe cardiac events (0.1%) and all cardiac events (0.3%) for the following year.
Comparison with recent publications
This study significantly expands our current knowledge of the prognostic value of CCTA. Min et al. (5) demonstrated a good correlation between CCTA results and all-cause mortality in a study of comparable size. The main limitation of this study is the missing distinction between cardiac and noncardiac deaths. Consequently, this study does not answer the question to what extent this correlation is influenced by concomitant noncardiac morbidity or by a possible selection bias, because CCTA is ordered more often for otherwise ill patients. Our observation that noncardiac deaths are far more common than cardiac deaths suggests that this confounding might be substantial. Consequently, the results of the Min et al. study (5) alone cannot be used as a base for decisions on further cardiac treatment.
Punziute et al. (6) and Gaemperli et al. (8) could demonstrate a significant prognostic value for cardiac events and a very good prognosis for patients without obstructive CAD. Limitations of both studies are their rather small study populations of 100 and 220 patients, respectively. Furthermore, patients with suspected and already known CAD were included in both studies, resulting in an inhomogeneous patient population. In addition, both studies included early revascularizations, which are predominantly driven by the CCTA itself, into their end points. Early revascularization procedures reflect the result of CCTA, but not long-term effects. Thus, we excluded these in the current analysis.
Gilard et al. (7) showed a low incidence of cardiac events after normal CCTA in patients with suspected CAD in a small study with 141 patients. The event rate of 0.6% per year for severe cardiac events was slightly higher than our event rates. No follow-up for patients with obstructive CAD for comparison was performed. Danciu et al. (9) investigated a very selected population of 421 patients with intermediate risk after scintigraphic myocardial perfusion stress imaging and showed a low event rate of 0.3% for patients assigned to medical treatment compared with 19% for the 28 patients referred to invasive angiography and not undergoing immediate revascularization.
Prognostic value of CCTA
One important goal of noninvasive imaging techniques for evaluation of cardiovascular risk is the ability to further differentiate high-risk patients from low-risk patients after conventional risk scoring. The current study demonstrates this ability for severe cardiac events and for all cardiac events. The significant difference in all cardiac events suggests that CCTA is not only capable of excluding obstructive CAD but also of assessing the risk of progression of CAD. Some studies showed that a considerable fraction of acute myocardial infarctions are caused by lesions classified as nonobstructive in previous angiograms (19,20). Nevertheless, in the current study, an extent of 50% lumen narrowing in visual assessment proved as an effective cutoff value for the prediction of cardiac events. A further evaluation of nonobstructive lesions regarding number, extent, or plaque composition was not performed because of the small number of events in the patient group without obstructive CAD.
Another important goal is the precise definition of a low-risk patient group. The established modalities are nuclear stress myocardial perfusion imaging and, more recently, magnetic resonance stress imaging, both consistently showing event rates for severe cardiac events below 1% per year in multiple studies (3,21,22). With a rate of 0.3% per year for all cardiac events and 0.1% per year for severe cardiac events in this large patient group with suspected CAD, the current study is an important step for establishing CCTA as an equivalent method. Obviously, the quoted weakness of the CCTA of only focusing on coronary anatomy and lumen narrowing and not assessing the extent of the resulting ischemia does not seem to play an important role in cardiac prognosis, at least in the first year, but certainly further evaluation of this topic is needed.
The negative predictive value of CCTA, like any other investigation, is strongly dependent on the pre-test risk. Because a good negative predictive value in a low-risk patient group can solely be caused by the low overall pre-test risk, we compared the observed event rate with the pre-test risk. The latter was estimated by the Framingham risk score, because it is widely accepted and well tested (15,16). The observed event rate among patients without obstructive CAD was significantly lower than predicted, whereas significantly more events occurred than predicted among patients with obstructive CAD. Thus, the current study demonstrates that CCTA appears to be very valuable in a further risk differentiation in addition to Framingham risk score.
Looking at the patients with obstructive CAD, nearly one-third of the patients advised to have invasive angiography did not undergo cardiac catheterization. Obviously, a significant number of patients were worried enough about the threat of cardiac disease to undergo a CT scan but were not worried enough to undergo invasive angiography, if necessary. Considering that all cardiac events occurred in patients with obstructive CAD by CCTA but not undergoing recommended invasive angiography emphasizes the importance of thorough patient information on the consequences of both a normal and an abnormal test result.
The overall event rates of 0.6% for severe cardiac events and 1.8% for all cardiac events are rather low. This finding precludes a valid multivariate analysis for comparison of CCTA with other risk predictors. In addition, the results of this study might not be valid for a high-risk patient population. The low event rate is, at least in part, a result of the high number of asymptomatic patients in the study population. Published data regarding the value of CCTA in this patient group are sparse at the moment.
The comparison of follow-up events between the 2 patient groups is influenced by further diagnostic and therapeutic procedures initiated by the investigation, first of all by subsequent coronary revascularizations. Although such effects are inevitable in a study observing routine clinical treatment pathways, one has to keep in mind that the result of the comparison probably is substantially influenced by how abnormal or equivocal results are handled. Nevertheless, the difference in the event rates was not influenced by complications of revascularization, because patients were censored for the end point of all cardiac events after the first revascularization (both early and late), and all severe cardiac events in the group with obstructive CAD occurred in patients who did not have recommended invasive angiography.
CCTA has emerged as a method for noninvasive assessment of CAD with an excellent negative predictive value, when compared with invasive coronary angiography. On top of this, our study clearly demonstrates the capability of CCTA for the prediction of cardiac events in patients with suspected CAD. Significantly more severe and all cardiac events occurred in patients with obstructive CAD by CCTA than in patients without obstructive CAD. The exclusion of obstructive CAD by CCTA is associated with a very high negative predictive value for cardiac events in the following 18 months, and the observed event rate is significantly lower than predicted by the Framingham risk score. These findings add substantial evidence to establishing CCTA as a routine modality for assessing the risk for cardiac events of patients with suspected CAD.
The authors are indebted to the medical and technical staff members of the computed tomography laboratory for their invaluable contribution, and to Mr. R. Distler for his assistance during follow-up.
This trial was supported in part by research grant KKF 05-05 from Deutsches Herzzentrum, Munich, Germany.
- Abbreviations and Acronyms
- coronary artery disease
- coronary computed tomography angiography
- confidence interval
- computed tomography
- Received October 3, 2008.
- Revision received November 6, 2008.
- Accepted November 12, 2008.
- American College of Cardiology Foundation
- Rosamond W.,
- Flegal K.,
- Furie K.,
- et al.
- Hachamovitch R.,
- Hayes S.,
- Friedman J.D.,
- et al.
- Jahnke C.,
- Nagel E.,
- Gebker R.,
- et al.
- Abdulla J.,
- Abildstrom S.Z.,
- Gotzsche O.,
- Christensen E.,
- Kober L.,
- TorpPedersen C.
- Min J.K.,
- Shaw L.J.,
- Devereux R.B.,
- et al.
- Pundziute G.,
- Schuijf J.D.,
- Jukema J.W.,
- et al.
- Danciu S.C.,
- Herrera C.J.,
- Stecy P.J.,
- Carell E.,
- Saltiel F.,
- Hines J.L.
- Morise A.P.,
- Jalisi F.
- Wilson P.W.,
- D'Agostino R.B.,
- Levy D.,
- Belanger A.M.,
- Silbershatz H.,
- Kannel W.B.
- Hausleiter J.,
- Meyer T.,
- Hadamitzky M.,
- et al.
- Kannel W.B.,
- Feinleib M.,
- McNamara P.M.,
- Garrison R.J.,
- Castelli W.P.
- Anderson K.M.,
- Wilson P.W.,
- Odell P.M.,
- Kannel W.B.
- R Development Core Team
- Ambrose J.A.,
- Winters S.L.,
- Arora R.R.,
- et al.
- Little W.C.,
- Constantinescu M.,
- Applegate R.J.,
- et al.
- Ingkanisorn W.P.,
- Kwong R.Y.,
- Bohme N.S.,
- et al.