Original Research

Reporting Coronary Artery Calcium on Low-Dose Computed Tomography Impacts Statin Management in a Lung Cancer Screening Population

Author and Disclosure Information

Background: Cigarette smoking is an independent risk factor for atherosclerotic cardiovascular disease (ASCVD). Concomitant use of low-dose computed tomography (LDCT) for coronary artery calcium (CAC) scoring with lung cancer screening (LCS) has been proposed to further determine ASCVD risk and mortality. We aimed to determine the validity of LDCT in identifying CAC and its impact on statin management.

Methods: We conducted a retrospective review from November 2020 to May 2021 of Military Health System (MHS) beneficiaries who received LCS with LDCT and were referred for CAC scoring with electrocardiogram-gated CT. Of the 190 participants initially identified, 170 met study eligibility. The Agatston method was used to score CAC on both scan types.

Results: Participants had a mean (SD) age of 62.1 (4.6) years and were 70.6% male. CAC was seen more on ECG-gated CT compared with LDCT (88% vs 74%, P < .001). The Spearman correlation and Kendall W coefficient of concordance of CAC scores between the 2 scan types was 0.945 ( P < .001) and 0.643, respectively. The κ statistic between CAC scores on the 2 different scans was 0.49 (SE κ = 0.048; 95% CI, -0.726-1.706), and the weighted κ statistic was 0.711. Bland-Altman analysis demonstrated a mean bias of 111.45 Agatston units, with limits of agreement between -268.64 and 491.54, suggesting CAC scores on electrocardiogram-gated CT were on average about 111 units higher than those on LDCT. There was a statistically significant proportion of nonstatin participants who met statin criteria based on additional CAC reporting ( P < .001).

Conclusions: CAC scores are highly correlated and concordant between LDCT and electrocardiogram-gated CT. Smokers undergoing annual LDCT may benefit from concomitant CAC scoring to help stratify ASCVD risk.



Cigarette smoking is an independent risk factor for lung cancer and atherosclerotic cardiovascular disease (ASCVD).1-3 The National Lung Screening Trial (NLST) demonstrated both lung cancer mortality reduction with the use of surveillance low-dose computed tomography (LDCT) and ASCVD as the most common cause of death among smokers.4,5 ASCVD remains the leading cause of death in the lung cancer screening (LCS) population.2,3 After publication of the NLST results, the US Preventive Services Task Force (USPSTF) established LCS eligibility among smokers and the Center for Medicare and Medicaid Services approved payment for annual LDCT in this group.1,6,7

Recently LDCT has been proposed as an adjunct diagnostic tool for detecting coronary artery calcium (CAC), which is independently associated with ASCVD and mortality.8-13 CAC scores have been recommended by the 2019 American College of Cardiology/American Heart Association cholesterol treatment guidelines and shown to be cost-effective in guiding statin therapy for patients with borderline to intermediate ASCVD risk.14-16 While CAC is conventionally quantified using electrocardiogram (ECG)-gated CT, these scans are not routinely performed in clinical practice because preventive CAC screening is neither recommended by the USPSTF nor covered by most insurance providers.17,18 LDCT, conversely, is reimbursable and a well-validated ASCVD risk predictor.18,19

In this study, we aimed to determine the validity of LDCT in identifying CAC among the military LCS population and whether it would impact statin recommendations based on 10-year ASCVD risk.


Participants were recruited from a retrospective cohort of 563 Military Health System (MHS) beneficiaries who received LCS with LDCT at Naval Medical Center Portsmouth (NMCP) in Virginia between January 1, 2019, and December 31, 2020. The 2013 USPSTF LCS guidelines were followed as the 2021 guidelines had not been published before the start of the study; thus, eligible participants included adults aged 55 to 80 years with at least a 30-pack-year smoking history and currently smoked or had quit within 15 years from the date of study consent.6,7

Between November 2020 and May 2021, study investigators screened 287 patient records and recruited 190 participants by telephone, starting with individuals who had the most recent LDCT and working backward until reaching the predetermined 170 subjects who had undergone in-office consents before ECG-gated CT scans. Since LDCT was not obtained simultaneously with the ECG-gated CT, participants were required to complete their gated CT within 24 months of their last LDCT. Of the 190 subjects initially recruited, those who were ineligible for LCS (n = 4), had a history of angioplasty, stent, or bypass revascularization procedure (n = 4), did not complete their ECG-gated CT within the specified time frame (n = 8), or withdrew from the study (n = 4) were excluded. While gated CT scans were scored for CAC in the present time, LDCT (previously only read for general lung pathology) was not scored until after participant consent. Patients were peripherally followed, via health record reviews, for 3 months after their gated CT to document any additional imaging ordered by their primary care practitioners. The study was approved by the NMCP Institutional Review Board.

Coronary Artery Calcification Scoring

We performed CT scans using Siemens SOMATOM Flash, a second-generation dual-source scanner; and GE LightSpeed VCT, a single-source, 64-slice scanner. A step-and-shoot prospective trigger technique was used, and contiguous axial images were reconstructed at 2.5-mm or 3-mm intervals for CAC quantification using the Agatston method.20 ECG-gated CT scans were electrocardiographically triggered at mid-diastole (70% of the R-R interval). Radiation dose reduction techniques involved adjustments of the mA according to body mass index and iterative reconstruction. LDCT scans were performed without ECG gating. We reconstructed contiguous axial images at 1-mm intervals for evaluation of the lung parenchyma. Similar dose-reduction techniques were used, to limit radiation exposure for each LDCT scan to < 1.5 mSv, per established guidelines.21 CAC on LDCT was also scored using the Agatston method. CAC was scored on the 2 scan types by different blinded reviewers.


We reviewed outpatient health records to obtain participants’ age, sex, medical history, statin use, smoking status (current or former), and pack-years. International Classification of Diseases, Tenth Revision codes within medical encounters were used to document prevalent hypertension, hyperlipidemia, and diabetes mellitus. Participants’ most recent low-density lipoprotein value (within 24 months of ECG-gated CT) was recorded and 10-year ASCVD risk scores were calculated using the pooled cohorts equation.

Statistical Analysis

A power analysis performed before study initiation determined that a prospective sample size of 170 would be sufficient to provide strength of correlation between CAC scores calculated from ECG-gated CT and LDCT and achieve a statistical power of at least 80%. The Wilcoxon rank sum and Fisher exact tests were used to evaluate differences in continuous and categorical CAC scores, respectively. Given skewed distributions, Spearman rank correlations and Kendall W coefficient of concordance were respectively used to evaluate correlation and concordance of CAC scores between the 2 scan types. κ statistics were used to rate agreement between categorical CAC scores. Bland-Altman analysis was performed to determine the bias and limits of agreement between ECG-gated CT and LDCT.22 For categorical CAC score analysis, participants were categorized into 5 groups according to standard Agatston score cut-off points. We defined the 5 categories of CAC for both scan types based on previous analysis from Rumberger and colleagues: CAC = 0 (absent), CAC = 1-10 (minimal), CAC = 11-100 (mild), CAC = 101-400 (moderate), CAC > 400 (severe).23 Of note, LDCT reports at NMCP include a visual CAC score using these qualitative descriptors that were available to LDCT reviewers. Analyses were conducted using SAS version 9.4 and Microsoft Excel; P values < .05 were considered statistically significant.


Recommended Reading