Coronary Artery Calcium Scoring in Clinical Trials
By Amit Mehta, MD, Vice President & Chief Medical Officer, Intrinsic Imaging, LLC
San Antonio (TX), Bolton (MA)
Volume 5, Issue 7
Coronary Artery Calcium scoring (CAC) is a well-established biomarker available in monitoring the effectiveness of therapies to treat Coronary Artery Disease (CAD) – the most frequent cause of death in developed nations. According to the American Heart Association, the total costs of all cardiovascular diseases in the United States are projected to be $616 billion in 2015. According to the American Heart Association, the total costs of all cardiovascular diseases in the United States are projected to be $616 billion in 2015.
As such, biopharmaceutical companies are spending considerable effort in developing novel therapies for the treatment of CAD. Of the multiple biomarkers available in monitoring the effectiveness of these therapies, Coronary Artery Calcium scoring is a well-established tool.
This article aims to provide guidance on the technical advances in Coronary Artery Calcium scoring and provide a discussion of issues key to success for Sponsors of clinical trials.
The technical advances in non-invasive imaging has allowed the development of screening methodologies to identify those individuals who are at higher risk of clinical manifestations of CAD. The cornerstone of prediction revolves around the evaluation of atherosclerotic plaque within the coronary arteries.
The pathophysiology of atherosclerosis evolves from changes in arterial lipoprotein to deposition of cholesterol in the arterial wall with associated inflammation and calcium deposition. The Agatston score, named after its developer Arthur Agatston, is a measure of calcium in the coronary vessels as measured on CT imaging. The original work was based on electron beam computed tomography (EBCT) which has now evolved in today’s imaging arena to high speed multi detector CT scanners (MDCT). The Agatston score is calculated using a weighted value assigned to the highest density calcification within the coronary artery being measured. The measurement is obtained in Hounsfield units and a score assigned based on area in square millimeters and a weighted score. The sum of these measurements is then calculated to give the total coronary artery calcium score. The CAC score is further refined by the introduction of lesion specific calcium-scoring which has been shown to be superior to the total score.
Because of the strong correlation between the presence of coronary calcification and underlying coronary atherosclerosis, CT imaging of coronary calcium has been determined to be a good predictor of cardiac events. A positive test is stratified into minimal (score less than 10), moderate (11 to 99), increased (100 to 400) or extensive (greater than 400) based on the amount of calcium. It is well established that patients with scores greater than 400 have an increased occurrence of coronary events.
Heterogeneity – Considerations in Clinical Trials
Recent work in the study of Coronary Artery Calcium scoring and heart disease has shown that greater plaque density appears to be relatively protective against future cardiovascular events. This is contrast to the existing body of knowledge and implications of coronary calcium measurement. Based on the large Multi Ethnic Study of Atherosclerosis (MESA), in more than 3000 men and women participants, it was found that patients who experienced a decreased number of adverse cardiac events over a seven year follow up period had denser calcified plaque. The patients with larger CAC volumes had more events related to coronary heart disease and cardiovascular disease while those with higher calcium density had fewer. This has brought into light the need for further study of the predictive value of Agatston scoring. The data from the MESA study has suggested that the Agatston area or volume scores alone are not optimal measures to use in CVD risk prediction.
Although further study is now underway to provide clinical trial Sponsors and clinicians optimal methodologies to study cardiovascular risk, the use of calcium scoring via various methods remains the mainstay of assessment. Thus, it is important that the quality of data being obtained is optimal for assessment and that imagers pay particular attention to the methodology utilized in acquiring data that is meaningful.
Evolution of Methods
A well-known shortcoming of the original quantification methods was limited reproducibility when repeated examinations were performed. This led to the evolution of new algorithms with the introduction of calcium mass and calcium volume scores to complement and possibly replace the traditional Agatston score. All of these methods have demonstrated reduction in inter- and intra-observer variability with the calcium mass score showing the best accuracy and reproducibility. The Agatston score, however, is still the most widely used methodology.
Originally, the Agatston score was based on the acquisition of 20 contiguous EBCT images with a 3 mm thickness. On each of the 20 images, calcifications were identified by setting a threshold of 130 Hounsfield units on structures greater than 1 mm2. A region of interest was placed on each lesion on each of the 20 images and the area and maximum CT number was determined. Due to the various limitations of this method including complexity, dependence on number of images, non-linearity in the amount of calcium to name several, the volume score methodology became more popular.
The volume score, as studies revealed, was more robust in terms of reproducibility. The volume score represents the volume of calcium calculated as the number of voxels in the volume of data that belongs to the calcium multiplied by the volume of one voxel. Again, a typical threshold of 130 Hounsfield units is utilized and all voxels that meet that attenuation value are utilized. This methodology, however, is very sensitive to partial volume averaging in CT imaging. Thus further work emerged in conjunction with improving scanner technology and the calcium mass score emerged as potentially the best measure of the amount of coronary artery calcium.
Calcium Mass Score
The calcium mass score is calculated by obtaining the mean CT number of the calcification shown on each CT image and then multiplying by the volume of the calcification on each image. The product obtained is proportional to the calcium mass on that image. Subsequently the total mass of calcium is obtained by summing all the independent scores.
Assessment of the Calcium Score Accuracy and the Precision of Measurement.
The reproducibility of Coronary Artery Calcium scoring measurements is of paramount importance in clinical trials where serial follow up examinations are conducted and represent the evolution of a disease process or the effect of a drug compound. The accuracy of measurement also is of importance when utilizing different scanners from different vendors at multiple clinical sites.
Willemink et al found in recent work that the Agatston scores differed substantially when CT scanners from various vendors were studied with simulations. This variation resulted in a change of cardiovascular risk classification in up to 6.5% of individuals to a high or low risk category when a CT scanner from a different vendor was used.1
This is troubling research for Sponsors conducting clinical trials who are acquiring data from multiple sites with multiple scanners. However, research has shown that in terms of the Agatston and volume scores for collected phantom Coronary Artery Calcium scoring data, the largest variation was in volume scores and less so for Agatston scores. This suggests that for widespread work, Agatston scores, with particular attention to standardization, will remain the primary imaging scoring tool given its familiarity and availability.
Given the noted variation between scanners, however, significant work is now underway in an effort to standardize CAC measurement from acquisition through post processing to obtain more accurate and precise data.
Standardization of Acquisition – Vendor Variance
In the early development of Coronary Artery Calcium scoring, it was evident by physicists that image reconstruction parameters and scoring parameters directly impacted outcome scores. In 2007, leaders in CAC technology worked to develop a consensus standard for quantification utilizing a multi-institutional, multi-manufacturer international consortium of cardiac radiologists, medical physicists and industry representatives.2 In this work, it became evident several parameters were essential to achieving reproducible results across various platforms. These included:
- The use of patient specific tube current to achieve a prescribed image noise
- The use of calcium mass score to eliminate scanner and patient size based variation
- The use of an anthropomorphic phantom containing calibration inserts and additional phantom rings to simulate small, medium and large patients
- Standardization of the calculation of the calcium mass score
By utilizing these recommended scanning parameters, it was noted the noise obtained is similar for small, medium and large phantoms for all MDCT vendors. The consensus also demonstrated that the use of a fixed calcium hydroxyapatite density threshold (100 mg/cm3) as compared with the use of a fixed CT number threshold of 130HU reduced interscanner variability in Agatston and calcium mass scores. Lastly, the standardized methods resulted in comparable spatial resolution and resulted in reduced radiation dose for small and medium sized patients. For clinical trial Sponsors and clinicians working in the acquisition of CAC data, it is crucial to pay attention to parameters at the acquisition level as noted above to ensure data obtained across patient populations is optimized for the post processing that ensues.
Post-Processing – Cross Platform Reproducibilty of Measurement
There are multiple software packages available on the market that are FDA approved for evaluation of MDCT and EBCT for Agatston scoring. As noted above, once parameters in acquisition are standardized, Coronary Artery Calcium scoring becomes more reproducible and gives more accurate measurements over time. However, while all scoring methods are sensitive to image reconstruction and scoring parameters, it has typically been thought that the software packages utilized for post-processing did not introduce any additional sensitivities. In recent work by Weininger et al, it was demonstrated that significant differences exist in the output of various software packages. 3 It was found in comparing three popular systems that each workstation produced different absolute numeric results for Agatston and volume scores for patient images that had been standardized across the three platforms. While this is initially troubling for large scale studies utilizing multiple sites, ultimately, the researchers concluded that although the systems produced different absolute numeric results, the statistical analysis showed good correlation between the workstations for both scoring methods and thus comparable CAC scoring results. The implication of these findings suggests that there is vendor independent reproducibility of coronary calcium scoring. In the clinical trial arena, however, it also suggests the necessity of utilizing standardized methodologies in imaging charters to obtain accurate and reproducible evaluation of CAC as a biomarker. Particularly relevant to trials, the investigators noted in their study that while the comparative analysis showed a high statistical correlation and comparability of CAC scores across software systems, a subgroup of the studied patients demonstrated divergences in risk group classification and thus Sponsors of clinical trials need to be cognizant of such possible discrepancies.
As noted, the use of Coronary Artery Calcium scoring in monitoring therapy or new drug development demands a keen understanding of standardization techniques to produce optimal accurate and reproducible results. As such, clinical trial Sponsors need to be acutely aware of the various issues facing CAC measurement and its utilization. When planning the inclusion of MDCT CAC in clinical trials, particular attention must be given not only to the particular methodology of calcium measurements employed (i.e. Agatston scores, volume scores or mass scores)but also to the standardization of acquisition parameters and post-processing of data to ensure optimal accuracy and reproducibility.
1. Willemink MJ, Vliegenthart R et al. Coronary Artery Calcification Scoring with State-of-the-Art CT Scanners from Different Vendors Has Substantial Effect on Risk Classification. Radiology: Volume 273: Number 3; 695-702
2. McCollough C, Ulzheimer S et al. Coronary Artery Calcium: A Multi-institutional, Multimanufacturer International Standard for Quantification at Cardiac CT. Radiology 2007 243:2, 527-538
3. Weininger M, Ritz K et al. Interplatform Reproducibility of CT Coronary Calcium Scoring Software. Radiology 2012; 265:70-77
4. Go AS et al. Heart Disease and Stroke Statistics-2014 Update: a report from the American Heart Association. Circulation. 2014. Published on line Dec 18, 2013, 10.1161/01. Cir0000441139,02102.80.
About the Author
Dr. Amit Mehta, MD, is the Vice President & Chief Medical Officer at Intrinsic Imaging. Dr. Amit Mehta completed his residency and fellowship training at Harvard Medical School at the prestigious Massachusetts General Hospital. He is double boarded and specialty certified in diagnostic radiology and vascular interventional radiology in Canada and the USA. He has widely published in radiology and served as a peer reviewer for Radiographics. He has authored three books in Radiology, published CD-ROMS for radiology education and has run several CME courses on computer applications in medicine and radiology. He is the past recipient of the George Brown and Radiological Society of North America awards for his research activities. He possesses a Certificate of Additional Qualification (CAQ) in Vascular and Interventional Radiology.