Coronary computed tomography angiography (CCTA) is a recently introduced noninvasive imaging modality that permits accurate detection and exclusion of coronary artery disease (CAD) (1- 3). Prior studies have examined the prognostic significance of CAD detection by CCTA, but have been generally limited to single centers and small cohorts (4- 14). Given the enormous evidence base on prognosis with other cardiac imaging modalities, the development of “real world” effectiveness data that are acquired across diverse healthcare settings and populations is a requisite criterion to guide appropriate application of CCTA. Fundamental to the use of CCTA is its ability to risk stratify younger and older women and men. To date, however, age- and sex-specific outcomes of CCTA-identified CAD have been absent or exploratory (15). To that end, we examined the predictive value of nonobstructive and obstructive CAD from a large cohort of 23,854 patients without known CAD for intermediate-term mortality risk and further investigated the relationship of mortality risk to CAD as stratified by age and sex.

CONFIRM (Coronary CT Angiography Evaluation for Clinical Outcomes: An International Multicenter Registry) enrolled consecutive adults ≥18 years of age between 2005 and 2009 who underwent ≥64-detector row CCTA for suspected CAD at 12 centers (Cedars Sinai Medical Center, Los Angeles, California; Harbor UCLA Medical Center, Los Angeles, California; Tennessee Heart and Vascular Institute, Hendersonville, Tennessee; Capital Cardiology Associates, Albany, New York; University of Munich, Munich, Germany; Ottawa Heart Institute, Ontario, Canada; Henry Ford Medical Center, Detroit, Michigan; Yonsei Medical Center, Seoul, Korea, University Hospital, Zurich, Switzerland; William Beaumont Hospital, Royal Oak, Michigan; Walter Reed Army Medical Center, Washington, DC; and University Hospital of Parma, Parma, Italy). Institutional review board approval was obtained at each center. Individuals with known CAD, as defined by prior myocardial infarction or coronary revascularization, were excluded from the present study analysis.

All patients were in normal sinus rhythm and were capable of the breath hold needed for CCTA. Patients with heart rates >70 beats/min were given oral or intravenous metoprolol as per local site protocol. All centers used intravenous metoprolol at the time of CCTA performance to lower heart rates below 70 beats/min. If the patient's heart rate did not drop below 70 beats/min, then CCTA was performed at the lowest heart rate.

Before the initiation of the scan, we prospectively collected information on the presence of categorical cardiac risk factors in each individual. Systemic arterial hypertension was defined as a documented history of high blood pressure or treatment with antihypertensive medications. Diabetes mellitus was defined by diagnosis of diabetes made previously by a physician and/or use of insulin or oral hypoglycemic agents. Dyslipidemia was defined as known but untreated dyslipidemia or current treatment with lipid-lowering medications. A positive smoking history was defined as current smoking or cessation of smoking within 3 months of testing. Family history of coronary heart disease was determined by patient query. Symptom presentation was classified into 1 of 4 categories: typical chest pain, atypical chest pain, noncardiac pain, or dyspnea (16).

The CCTA scans were performed on a variety of different scanner platforms (Lightspeed VCT, GE Healthcare, Milwaukee, Wisconsin; Somatom Definition CT, Siemens, Erlangen, Germany; Somatom Definition Flash CT, Siemens). Imaging of a test bolus of contrast was performed at 2 mm superior to the take-off of the left main coronary artery for precise timing of contrast injection. During the CCTA acquisition, 80 to 140 ml of iodinated contrast (Isovue 370, Bracco Diagnostics, Princeton, New Jersey; Omnipaque, GE Healthcare, Princeton, New Jersey; Visipaque, GE Healthcare, Princeton, New Jersey; or Imeron 350; Bracco Atlana Pharma, Konstanz, Germany) was injected, followed by a 50-ml saline flush. Contrast timing was performed to optimize uniform contrast enhancement of the coronary arteries. The scan parameters were as follows: 64 × 0.625/0.750-mm collimation, tube voltage 100 or 120 mV, effective 400 to 650 mA. Dose reduction strategies—including electrocardiogram-gated tube current modulation, reduced tube voltage, and prospective axial triggering—were used whenever feasible. Estimated radiation dose for CCTA ranged from 3 to 18 mSv.

Helical or axial scan data were obtained with retrospective or prospective electrocardiogram gating, respectively. Images were reconstructed immediately after completion of the scan to identify motion-free coronary artery images. Optimal phase reconstruction was assessed by comparison of different phases, if available, and the phase with the least amount of coronary artery motion was chosen for analysis. Multiple phases were utilized for image interpretation if minimal coronary artery motion was different for different arteries. CCTAs were evaluated by an array of post-processing imaging techniques, including axial, multiplanar reformat, maximum-intensity projection, and short-axis cross-sectional views. In all individuals, irrespective of image quality, every arterial segment was scored in an intent-to-diagnose fashion. If a coronary artery segment was uninterpretable despite these multiple techniques, the nonevaluable segment was scored similarly to the most proximal segment that was evaluable.

All scans were analyzed by Level III–equivalent cardiologists with experience interpreting several thousand CCTA scans. Interpretation of CCTA was uniform across all study sites, with coronary segments visually scored for the presence of coronary plaque using a 16-segment coronary artery model in an intent-to-diagnose fashion. In each coronary artery segment, coronary atherosclerosis was defined as tissue structures >1 mm² that existed either within the coronary artery lumen or adjacent to the coronary artery lumen that could be discriminated from surrounding pericardial tissue, epicardial fat, or the vessel lumen itself. Coronary atherosclerotic lesions were quantified for stenosis by visual estimation. Luminal-diameter stenosis severity was scored as none (0% luminal stenosis), mild (1% to 49% luminal stenosis), moderate (50% to 69% luminal stenosis), or severe (≥70% luminal stenosis). Percent obstruction of coronary artery lumen was based on a comparison of the luminal diameter of the segment exhibiting obstruction to the luminal diameter of the most normal-appearing site immediately proximal to the plaque. In instances in which plaque was highly calcified, 2-dimensional oblique images were also visualized without maximal-intensity projection (i.e., 0.625- to 0.75-mm isotropic voxel resolution) or multiplanar reformats with cross-sectional views to minimize partial-volume averaging artifact of calcium.

Plaque severity was graded on a per-patient, per-vessel, and per-segment level. Per-patient maximal plaque severity was defined by the maximal intraluminal stenosis in any of the coronary segments at the ≥50% stenosis or ≥70% stenosis threshold.

For purposes of classification for per-vessel analyses, we considered 4 arterial territories: left main artery (LM), left anterior descending artery (LAD), left circumflex artery (LCx), and right coronary artery (RCA). Obstructive CAD in the diagonal branches, obtuse marginal branches, and posterolateral branches was considered as part of the LAD, LCx, and RCA system, respectively. The posterior descending artery was considered as part of the RCA or LCx system, depending on the coronary artery dominance. A ≥50% stenosis in the LM was considered obstructive in all models. Per-vessel CAD severity was defined by ≥50% stenosis or ≥70% stenosis in 0, 1, 2, or 3 coronary artery vessels.

Per-segment analysis was judged for individual coronary artery segments that included a 16-segment model, as we have previously described (4). Similarly, the incremental hazards of CAD for increasing numbers of segments were calculated as clinical coronary artery plaque scores, as we have previously described (4). A segment involvement score was calculated as the total number of coronary artery segments exhibiting plaque, irrespective of the degree of luminal stenosis within each segment (minimum = 0; maximum = 16). A segment stenosis score was used as a measure of overall coronary artery plaque extent. Each individual coronary segment was graded as having no to severe plaque (i.e., scores from 0 to 3) based on extent of obstruction of coronary luminal diameter. Then the extent scores of all 16 individual segments were summed to yield a total score ranging from 0 to 48. We further examined risk in association with any severe proximal stenosis in the LAD, LCx, or RCA vessels. Finally, we examined risk for any plaque within the LM.

The primary endpoint was time to death from all causes. Follow-up procedures were approved by all study centers' institutional review boards. Death status for non-U.S. centers was gathered by clinical visits, telephone contacts, and questionnaires sent by mail, with verification of all reported events by hospital records or direct contact with a patient's attending physician. Death status for U.S. centers was ascertained either by query of the Social Security Death Index or by scripted interview by experienced physician and/or nurse study investigators.

SPSS version 12.0 (SPSS Inc., Chicago, Illinois) and SAS version 9.2 (SAS Institute, Cary, North Carolina) were used for all statistical analyses. Categorical variables are presented as frequencies and continuous variables as mean ± SD. Variables were compared with chi-square statistic for categorical variables and by Student unpaired t test for continuous variables. Time to death from all causes and death rates were calculated using univariable Cox proportional hazards models. In each case, the proportional hazards assumption was met. Adjusted models were also devised including multivariable stepwise models adjusting for baseline demographics, cardiac risk factors, typicality of angina, and pre-test likelihood of obstructive CAD. Adjusted models were also developed to test first-order interactions related to age, sex, and study site. A hazard ratio (HR) and 95% confidence interval (CI) were calculated from the Cox models. A 2-tailed p < 0.05 was considered statistically significant.

Amongst 27,125 consecutive patients undergoing CCTA at 12 centers for whom per-segment CAD data were available, 2,350 patients with a history of myocardial infarction, coronary revascularization, and cardiac transplant were excluded. The final analysis cohort consisted of 24,775 patients. Follow-up was obtained for 23,854 patients (96.3%), with 921 patients lost to follow-up. The study cohort was middle-aged (age 57 ± 13 years, 54% male) with a high prevalence of cardiovascular risk factors and symptoms. They presented with typical or atypical angina in the majority of cases, with the majority of individuals having intermediate or high pre-test likelihood of obstructive CAD. Excluded patients had a higher pre-test likelihood of CAD (Table 1).

Survival was examined after a mean follow-up of 2.3 ± 1.1 years (median 2.1 years; interquartile range: 1.5 to 3.1 years), at which point 404 deaths were recorded. Increasing severity of CAD was associated with male sex, diabetes, hypertension, dyslipidemia, family history of CAD, current smoking, typical angina, and high pre-test likelihood of CAD (p < 0.0001 for all) (Table 2). In univariable Cox proportional hazards models, increased hazard for death was associated with advanced age, diabetes, hypertension, untreated dyslipidemia, and current smoking, but not family history of CAD or pre-test CAD likelihood (Table 3).

As compared with patients who were alive at follow-up, patients who died had significantly more severe coronary artery stenoses in the majority of coronary segments (Table 4).

In both univariable as well as multivariable Cox regression analysis considering age and CAD risk factors, all-cause mortality was predicted by maximal per-patient nonobstructive and obstructive CAD, whether using a definition of obstructive CAD as 1% to 49% or 1% to 69% stenosis ((Table 5),Figure 1).

By both univariable and multivariable Cox models, per-vessel assessment of obstructive CAD demonstrated a dose-response relationship for increased hazards for death for 1-vessel, 2-vessel, 3-vessel, or LM CAD ((Table 5), Figure 2). Similarly, in both univariable and multivariable Cox regression analysis, on a per-segment basis, higher rates of mortality were associated with greater numbers of segments with plaque, with stenosis-adjusted segments with plaque, with any severe proximal stenosis, and with any plaque within the LM (Table 5).

Sixty-six deaths (0.65%) occurred in patients without evident CAD by CCTA (n = 10,146; 43%), resulting in a mean annualized death rate of 0.28%. In additional analyses of patients undergoing CCTA followed for ≥4 years (n = 1,816) without evidence by CCTA (n = 1,009), annualized death rates were 0.22%.

Individuals <65 years of age had lower pre-test probability of CAD than those ≥65 years of age (pre-test probability low 34% vs. 16%; intermediate 60% vs. 67%, high 6% vs. 17%, chi-square p < 0.0001). As compared with individuals without CAD within respective age groups, patients <65 years of age experienced higher hazards for mortality if 2- or 3-vessel/LM obstructive CAD was present than patients ≥65 years of age, with similar rates of death for nonobstructive and 1-vessel obstructive CAD ((Table 6),Figure 3).

Female patients referred to CCTA had higher pre-test probability of CAD than men. Differences in pre-test likelihood for obstructive CAD existed by sex, with male and female patients presenting with low (pre-test probability low 35% vs. 24%, intermediate 56% vs. 68%, and high 9% vs. 9%; p < 0.0001). As compared with individuals without CAD within respective sexes, women experienced higher hazards for mortality for 3-vessel or LM obstructive CAD than males, with similar rates of death for nonobstructive, 1-vessel, and 2-vessel obstructive CAD ((Table 7),Figure 4).

When stratified by both age and sex, differences in multivariable risk-adjusted hazards for mortality were observed for nonobstructive and 1-, 2-, and 3-vessel or LM obstructive CAD (Table 8). Tests for interactions of age and CAD (p = 0.21) and sex and CAD (p = 0.76) did not reveal significant relationships.

These results of the CONFIRM registry represent the first prospective international multicenter data to relate CCTA-determined extent and severity of CAD to all-cause mortality and demonstrate the independent prognostic value of both obstructive as well as nonobstructive CAD by CCTA. Importantly, this study had adequate sample size and was adequately powered (beta >0.90, alpha <0.001) to permit differential risk stratification of individuals as categorized by age group and sex. The findings of this study should be considered widely generalizable, given the high number of enrolled subjects; the inclusion of numerous clinical sites within North America, Europe, and Asia; and the uniformity of study results across sites.

Despite increasing adoption of CCTA for clinical use in individuals with suspected CAD, limited “real world” effectiveness evidence still exists to support the prognostic significance of CAD findings as detected by current-generation ≥64-detector row CCTA. Furthermore, prior studies to date that have examined the ability of CCTA findings to stratify risk have been generally limited to single centers with relatively small sample sizes. We previously reported the predictive value of CCTA measures of CAD extent and severity in a 2-center study of 5,330 consecutive patients, but data were limited in that population to vessel-based analyses, and the prognostic potential of segment-based as well as nonobstructive CAD detection by CCTA could not be examined (5). In a separate analysis, we reported the prognostic significance of CCTA-identified CAD findings in 1,256 consecutive patients with suspected CAD for the prediction of major adverse cardiac events (8). Although we noted a 16- to 17-fold increased risk in myocardial events in patients with obstructive CAD, the observed event rates were low (0.6% to 1.8%) and the follow-up duration shorter than in the present study. In this regard, the current data extend prior studies by examining CCTA findings of CAD in a large consecutive cohort comprising multiple international sites that was evaluated by current-generation computed tomography (CT) technology for measures of both obstructive and nonobstructive CAD. That CCTA can effectively risk stratify individuals without known CAD should be invaluable for guiding the development of clinical practice guidelines and appropriate use criteria.

One notable finding in our study was that although the pre-test likelihood of obstructive CAD—as estimated by the Diamond-Forrester (D-F) tabular method—was highly predictive of the presence of obstructive CAD by CCTA, it did not demonstrate predictive value for incident death in this large cohort of patients with suspected CAD. At present, clinicians commonly use D-F pre-test estimations of likelihood of obstructive CAD as a metric to determine whether patients would benefit from noninvasive testing. Because global clinical risk scores are currently lacking for symptomatic stable individuals with suspected CAD, many have also adopted use of the D-F method for “risk” assessment (17- 18). The current data suggest that use of the D-F method is insufficient for this purpose and should not be used solely as a judge of risk.

We also identified a utility of nonobstructive CAD detection for risk stratification of individuals at heightened risk of incident death. Indeed, individuals with nonobstructive CAD (HR: 1.62; 95% CI: 1.08 to 2.43) by CCTA experienced a mortality risk that was similar to that of those with obstructive 1-vessel CAD (HR: 1.75; 95% CI: 1.12 to 2.72). These data corroborate our prior studies using older generation electron beam CT technology, where the presence of nonobstructive CAD conferred similar risk as 1-vessel obstructive CAD. These findings have important implications, as patients with nonobstructive CAD comprise the majority of patients who experience myocardial events and for whom functional stress testing aimed at detecting flow-limiting coronary artery stenoses would be expectedly negative (19). Given a robust evidence base demonstrating the highly salutatory effect of primary prevention of CAD events in at-risk individuals, future studies should be performed to determine the effect of aggressive medical therapy and lifestyle modification for individuals with CCTA-identified nonobstructive CAD.

We recently reported the results of an exploratory analysis of 1,127 consecutive patients with suspected CAD undergoing 16-detector row CCTA identified to have only obstructive CAD. In this study, 490 patients were identified as having nonobstructive CAD (as defined by maximal <50% luminal diameter stenosis at the per-patient level) (15). In 4-year follow-up, the number of coronary segments exhibiting nonobstructive CAD was predictive of incident death in women but not in men. This study expands on these prior published results by using a study cohort of sufficient magnitude to examine the presence of nonobstructive CAD at the per-patient level. The present data corroborate the prior data, revealing a prognostic value for per-patient nonobstructive CAD detection in women but not in men. Further, the current study extends prior study results by demonstration that 1-vessel obstructive CAD also confers heightened risk of death in women but not in men. Numerous possibilities exist to explain these findings. Women in our study had lower rates of both obstructive and nonobstructive CAD in contrast to their male counterparts, despite higher estimated pre-test probability. This generally lower prevalence of disease in women has been historically associated with lower rates of invasive coronary angiographic evaluation and often leads to “exclusion” of cardiac causes of symptoms in women, despite a higher rate of hospitalization for angina as compared with men (20- 21). It remains possible in this open-label study that the identification of nonobstructive disease in women led clinicians to pursue alternative noncardiac diagnoses for symptoms, and lack of aggressive treatment for these CAD findings resulted in heightened risk of incident death. Future studies should carefully evaluate this potential explanation and should determine the effect of primary prevention with aggressive medical therapy in this cohort.

In the present study, we also observed differential risk stratification of CCTA-identified CAD findings by age groups. Although risk of all-cause death increased in a generally linear fashion for patients ≥65 years of age for extent and severity of CAD, patients <65 years of age experienced a more abrupt increase in risk of death associated with 2- and 3-vessel or LM CAD over nonobstructive or 1-vessel CAD. Although numerous explanations exist to account for these findings, it may be that younger patients with greater extent and severity of CAD represent a cohort with more aggressive forms of atherosclerosis than their older counterparts, thus resulting in a higher risk than for older patients with more insidious atherosclerosis.

Finally, we observed a very low rate of death for individuals without evident CAD by CCTA. This low rate of death validates the favorable prognosis that has been uniformly observed in prior smaller registries and emphasizes a clinical value of CCTA for identification of individuals in whom no further additional testing and/or therapy is necessary or indicated (4- 14). Using older generation electron beam CT technology, Ostrom et al. (22) demonstrated that this very low death rate continues to persist for up to 7 years from the time of the CCTA, and the present results are in direct accordance with those findings. Similarly, the long-term ≥4-year prognosis of patients in the present registry without evident CAD by CCTA was extremely favorable, with a 0.22% annualized death rate. These results may inform clinicians on the need for repeat testing in patients with normal CCTA and suggest a “warranty” period of a normal CCTA to last at least 4 years.

Although this study addresses many of the shortcomings of prior analyses examining the prognostic value of CCTA, it is not without limitations. For the present analysis, the major endpoint was all-cause mortality. Other “softer” endpoints—including myocardial infarction, unstable angina, or CAD-related hospitalization—were not included in this initial analysis. Although use of all-cause death mitigates ascertainment bias, it will nevertheless be important to perform future studies examining the risk of major adverse cardiac events in relation to CCTA findings. Further, referral bias due to excluded patients and treatment of individuals based on CCTA findings of CAD are unknown in this open-label multicenter registry. Whether percutaneous or surgical coronary revascularization, enhanced medical therapy, or lifestyle modifications occurred after CCTA performance is unknown. We have previously demonstrated that CAD risk factor control is improved in direct relation to the extent and severity of CCTA CAD findings, but whether this affects mortality has not been explored. For proper evaluation of this issue, large-scale trials with prescribed treatment algorithms will be necessary. In addition, our analysis entailed the evaluation of CAD by semiquantitative visual analysis rather than by volumetric quantification of plaque. At the time of initiation of the study (and to date), no automated validated software existed for automatic quantification of plaque or stenosis severity. Instead, this multicenter study used experienced level III CCTA imagers and used a uniform grading system that is most commonly used in daily clinical practice. Finally, this study examined only patients without history of known CAD. As such, whether these findings can be extrapolated to those with prior myocardial infarction or coronary revascularization needs to be tested in future studies.

In the large, prospective, international, multicenter CONFIRM registry, extent and severity of CAD by CCTA successfully identifies individuals at heightened risk for all-cause mortality. Presence of both obstructive and nonobstructive CAD by CCTA on a per-patient, per-vessel, and per-segment basis portends worsened prognosis, with differential risk noted between sex and age groups. Importantly, individuals without evident CAD by CCTA are at very low risk of death.