Thoracic aortic calcium(TAC) is an important marker of extracoronary atherosclerosis with established predictive value for all-cause mortality. We sought to explore the predictive value of TAC for stroke mortality, independent of the more established coronary artery calcium (CAC) score. The CAC Consortium is a retrospectively assembled database of 66,636 patients aged ≥18 years with no previous history of cardiovascular disease, baseline CAC scans for risk stratification, and follow-up for 12 ± 4 years. CAC scans capture the adjacent thoracic aorta, enabling assessment of TAC from the same images. TAC was available in 41,066 (62%), and was primarily analyzed as present or not present. To account for competing risks for nonstroke death, we utilized multivariable-adjusted Fine and Gray competing risk regression models adjusted for traditional cardiovascular risk factors and CAC score. The mean age of participants was 53.8 ± 10.3 years, with 34.4% female. There were 110 stroke deaths during follow-up. The unadjusted subdistribution hazard ratio (SHR) for stroke mortality in those who had TAC present compared with those who did not was 8.80 (95% confidence interval [CI]: 5.97, 12.98). After adjusting for traditional risk factors and CAC score, the SHR was 2.21 (95% CI:1.39,3.49). In sex-stratified analyses, the fully adjusted SHR for females was 3.42 (95% CI: 1.74, 6.73) while for males it was 1.55 (95% CI: 0.83, 2.90). TAC was associated with stroke mortality independent of CAC and traditional risk factors, more so in women. The presence of TAC appears to be an independent risk marker for stroke mortality.
Thoracic aortic calcium (TAC), the most common form of extracoronary calcification, is an important marker of systemic atherosclerosis with known predictive value for all-cause mortality and 10-year coronary heart disease (CHD) risk, including when used in addition to the more established Agatston Coronary Artery Calcium (CAC) score. It can be measured with the same cardiac CT scans used to assess CAC, allowing its assessment without additional exposure to radiation and at no additional cost. Furthermore, TAC can be accurately assessed on all chest CT scans regardless of ECG gating. Although CAC and other extracoronary calcification have been associated with the incidence of stroke, , there is minimal information on the utility of TAC in estimating stroke outcomes. TAC has been associated with increased inflammatory markers, as well as increased risk of stroke and other cardiovascular events. , Additionally, the aorta is the main conduit large artery in the body, and is more directly connected to the carotid arteries which provide the bulk of the blood supply of the brain. Thus, TAC might be a better predictor of stroke outcomes than CAC, which is at best, a modest predictor of stroke. We therefore sought to explore the predictive value of TAC for stroke mortality independent of the CAC score. We also explored the association between TAC and stroke mortality by sex.
Methods
The CAC Consortium is a retrospectively assembled database of 66,636 patients aged ≥18 years with no previous history of cardiovascular disease (CVD), who had CAC scans done for risk stratification, with long-term follow-up of 12 ± 4 years for mortality outcomes. Participants were recruited from four institutions in the United States: Cedars-Sinai Medical Center, Los Angeles, CA; PrevaHealth Wellness Diagnostic Center, Colombus, OH; Harbor-UCLA Medical Center, Torrance, CA; and Minneapolis Heart Institute, Minneapolis, MN. Consent was obtained from all participants at their participation site and IRB approval was obtained from the Johns Hopkins University School of Medicine for coordinating center activities and death ascertainment.
For this study, we restricted our analysis to 41,066 participants with available information on the presence or absence of TAC.
ECG-gated noncontrast scans were performed for each participant using electron bean tomography or multidetector CT. A standard protocol was adapted for each scanner technology used. CAC scans capture a view of the adjacent thoracic aorta, enabling assessment of TAC at no extra cost. TAC was primarily assessed as present or absent. In addition, TAC score was assessed using the Agatston score in 34,024 patients (83% of the total analytical sample). In this subgroup, TAC area was calculated as TAC volume score/2.5 (slice thickness of multidetector CT scanners), and peak TAC density was calculated as TAC score/TAC area.
Patient demographic information, risk factor information and laboratory data were collected during routine clinical visit or at the time of the CAC scan. Hypertension was defined as a history of clinical diagnosis of hypertension or hypertensive medication therapy. Diabetes was defined as a previous clinical diagnosis of diabetes or management on oral hypoglycemic drugs or insulin. Dyslipidemia was defined as a history of clinical diagnosis of primary hyperlipidemia, dyslipidemia, management on any lipid lowering therapy or laboratory data showing LDL-C >160 mg/dl, HDL-C <40 mg/dl in men or <50 mg/dl in women, or fasting triglycerides>150 mg/dl. Smoking status was classified as current, former or never smoking. Family history of CHD was defined as a first degree relative with a history of CHD. CAC groups were defined as CAC = 0, CAC = 1–99, CAC = 100–399, CAC ≥400. Multiple imputation was used to account for any partially missing risk factor data. Mortality status was determined from the Social Security Index Death Master File, and death certificates were obtained from the National Death Index. Causes of death were then categorized using the International Classification of Diseases. The primary outcome of interest in our study was stroke mortality (listed as the primary underlying cause of death).
Demographic characteristics were summarized by TAC category (present/absent). Means and proportions were reported for continuous and categorical data respectively. Differences between categorical variables were tested using chi square statistic, while differences between continuous variables were tested using 2 sample t test. The cumulative incidence for stroke mortality by TAC and CAC groups was assessed and cumulative incidence function curves were derived and plotted. To assess the predictive value of TAC for stroke mortality while accounting for competing risks for death from other causes, we utilized Fine and Gray’s competing risk regression model. We report the relationship between TAC and stroke mortality risk with subdistribution hazard ratios (SHR) and 95% CI, using four models outlined as follows: Model 1 was unadjusted, Model 2 was adjusted for age and sex, Model 3 was defined as Model 2 plus traditional cardiovascular risk factors (hypertension, hyperlipidemia, cigarette smoking, diabetes, family history of CHD), Model 4 was defined as Model 3 plus CAC score groups.
Based on our findings and a priori knowledge of the strong effect of sex on TAC and outcomes, the models were further stratified by sex. We further tested for an effect modification of sex on the relationship between TAC and stroke mortality. Finally, to further assess the discriminatory value of TAC independent of CAC for the prediction of stroke mortality, we assessed the area under reviewer operating curves for fully-adjusted models with and without TAC.
Unlike TAC presence or TAC score, previous research found TAC density to be protective against CHD and CVD events, after accounting for TAC area or volume. , Thus, in further exploratory analysis within the subset with TAC scores, we assessed the relationship between peak density TAC and stroke mortality.
All analyses were run with Stata model 15.1 SE.
Results
Overall, the mean age of participants was 53.8 ± 10 years and 34.4% were female. Among the 41,066 participants studied, 11,684(28.5%) had TAC. Compared with individuals without TAC, those with TAC were older and more likely to be female, have hypertension, dyslipidemia, diabetes, higher CAC categories, smoke cigarettes, and have higher atherosclerotic cardiovascular disease risk scores ( Table 1 ).
| Characteristic | Thoracic aortic calcium | |
|---|---|---|
| YES (n = 11,684) | NO (n = 29,382) | |
| Age (years) (Mean± S.D) | 61.8 ± 9.6 | 50.7 ± 8.7 |
| Women | 4,340 (37.1%) | 9,776 (33.3%) |
| Hypertension | 5,176 (44.3%) | 7,183 (24.5%) |
| Dyslipidemia * | 7,101 (60.8%) | 14,039 (47.8%) |
| Diabetes mellitus | 1,190 (10.2%) | 1,313 (4.5%) |
| Smokers | 1,294 (11.1%) | 2,733 (9.3%) |
| Family history of CHD | 4,868 (41.7%) | 12,570 (42.8%) |
| CAC score group | ||
| 0 | 2,286 (19.6%) | 17,190 (58.5%) |
| 1-100 | 3,702 (31.7%) | 8,225 (28.0%) |
| 100-400 | 2,762 (23.6%) | 2,653 (9.0%) |
| >400 | 2,934 (25.1%) | 1,314 (4.5%) |
| ASCVD risk categories | ||
| <5% | 3,380 (28.9%) | 20,844 (70.9%) |
| 5-7% | 1,673 (14.3%) | 3,701(12.6%) |
| >7.5% | 6,631 (56.8%) | 4,837 (16.5%) |
⁎ defined as a history of clinical diagnosis of primary hyperlipidemia, dyslipidemia, management on any lipid lowering therapy or laboratory data showing LDL-C >160 mg/dl, HDL-C <40mg/dl in men or <50 mg/dl in women, or fasting triglycerides >150 mg/dl.
Over a mean follow-up of 10.7 ± 3.0 years, 143 individuals had stroke mortality, of whom 110 (76.9%) had TAC present. At the end of follow-up, the cumulative incidence of stroke mortality for individuals with TAC present was 1.2%, while those without TAC had a cumulative incidence of 0.1% ( Figure 1 ). Among individuals with TAC present, the cumulative incidence of stroke deaths in those with CAC scores of 0, 1–99, 100–399 and ≥400 was 0.6%, 0.7%, 1.0%, and 2.4% respectively, compared with individuals without TAC who had cumulative incidence estimates of 0.1%, 0.2%, 0.1%, and 0.3% for the respective CAC categories ( Figure 2 ).
