| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Science to Practice |
Department of Radiology,
Stanford University School of Medicine,
805 Tolman Dr,
Stanford, CA 94305 ,
lwexler{at}stanford.edu
SUMMARY
Multiple technical factors are employed to improve the accuracy and reproducibility of measurements of coronary artery calcification at computed tomography (CT). Rutten et al (1) have demonstrated that an important factor affecting measurement variability is the partial volume effect that operates when contiguous, rather than overlapping, scans are reconstructed. The practical implication of this error might be a false-negative study or inaccurate categorization of the severity of a patient's risk. The authors propose that manufacturers offer the ability to construct overlapping scans with multidetector CT scanners from prospectively electrocardiographically-triggered examinations obtained for quantifying coronary calcium.
THE SETTING
In a recent review (2), emphasis was placed on the fact that coronary calcium is always associated with atherosclerotic burden and indicates a stage in the healing process of plaque rupture. The predictive value of a coronary calcium measurement has been determined in a number of clinical circumstances (eg, its negative predictive value in patients presenting with acute chest pain and its potential to help refine the Framingham risk score to more accurately stratify risk for a future coronary event or mortality, thereby helping to determine risk intervention therapy). However, for a test to be accepted and useful, its results should be reproducible. This is particularly true if results will be followed over time to assess progression or regression of disease. In this context, the report by Rutten et al (1) in this issue of Radiology describes how a small variation in the starting position of the scan can have a substantial effect on whether a patient has a positive score for coronary calcium (>0 vs 0) and can account for a change in risk stratification category in some patients.
|
It is challenging to image calcifications in the wall of coronary arteries because of their small size and constant motion during the cardiac cycle. Shortened scanning times, improved signal-to-noise ratio, electrocardiographic (ECG) gating, and selection of section thickness and threshold values have contributed to improved reproducibility, although interscan variability remains high whether electron-beam CT or multidetector CT is performed (3). Increased reproducibility is gained by using rigorous protocols, automated image analysis, and a standardized phantom for comparison (4). Rutten et al (1) have demonstrated that an important factor undermining reproducibility is the use of contiguous, rather than overlapping, section reconstruction. Because of the dimensions of the voxel, the partial volume effect will result in partial filling of some voxels with calcium, resulting in a CT number that falls below the threshold. The effect will be greatest for the lowest values when a patient might have a false-negative score because none of the voxels exceed the threshold for calcium. The actual variation in CT number was greater at higher calcium values, but the percentage change was less, thus causing fewer changes in risk group.
The Agatston score (5) is a nonlinear variable, whereas the measurements of the mass and volume of calcium is a continuous variable and is less affected by many of the factors that determine reproducibility (6). Rutten et al (1) showed that the partial volume effect affected mass and volume estimates to a lesser extent than the Agatston score for calculating coronary calcium.
THE PRACTICE
Clinical use: Rutten et al (1) studied women at relatively low risk for coronary calcifications, a group with important differences from other populations of interest. For example, patients presenting with acute chest pain are in a higher risk group. For them, a negative calcium score or measurement will be of greater importance than for a group of low-risk patients, and the partial volume effect might thus account for many of the less than 5% false-negative studies in such patients. A more widely applicable use of the coronary calcium measurement would be to improve risk stratification in patients at intermediate risk for a coronary event on the basis of their Framingham risk score (7). Patients at high risk (10-year risk for a hard cardiac event, either death or myocardial infarction, >20%) according to the Framingham risk score are recommended to receive aggressive risk modifications and drug therapy, while those in the low-risk category (10-year risk for hard cardiac event, <10%) are advised to follow best practices for risk reduction. Those in the intermediate ranges of moderate or moderately high risk might be restratified into a lower risk group or, more likely, into a high-risk group, thus affecting their recommended treatment. The number of patients in these categories in the study of Rutten et al was small. About 8%–10% of patients were reclassified into a higher or lower category because of simulated repositioning, which suggests that their study should be repeated on populations that are at greater risk.
Future opportunities and challenges: To use changes in coronary calcium as a measure of the effectiveness of drug therapy or other interventions, it is important to know the reproducibility of the measurement and whether the starting value was low or high. The study of Rutten et al (1) had small numbers of patients in all categories. Other study results have shown between 10%–25% interscan variability (8). Some of this variability can now be ascribed to the partial volume effect on the basis of the work of Rutten et al. Rutten et al (1) recommend that the optimal solution for overcoming the limitation due to partial volume effect would be to reconstruct overlapping data sets from the raw data acquired with multidetector CT with prospective ECG triggering. Prospective triggering uses less radiation than retrospective gating, and the overlapping sections reduce the influence of partial volume effects. Further studies are indicated to see whether overlapping images from multidetector CT will further reduce interscan variability. In the meantime, manufacturers should be encouraged to offer this potential improvement on their multidetector CT scanners.
FOOTNOTES
See also the article by Rutten et al in this issue.
References
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
