Objective: The potential clinical impact of variability in the measurement of coronary artery calcification with sequential MDCT was evaluated using Agatston, volume, and mass scoring algorithms.
Subjects and methods: Fifty-six patients were imaged twice using an identical prospectively ECG-triggered sequential scanning protocol. The Agatston, volume, and mass scores were computed by two observers independently. In addition, a patient's total Agatston score was referenced to an age- and sex-stratified database to determine a percentile ranking. Interscan, interobserver, and intraobserver variability and the resultant impact on patients' risk stratifications were assessed.
Results: Significant interscan differences were found for all mean coronary calcium scores (Wilcoxson's signed rank test, p <0.0001). Although the median percentage of interscan variability was low for all scoring methods, the interquartile range was wide, indicating significant variability in the data. Median scores (lower quartile-upper quartile) for observers 1 and 2, respectively, were as follows: Agatston, 5% (0-79%) and 6% (0-83%); volume, 12% (0-51%) and 12% (0-57%); and mass, 14% (0-57%) and 14% (0-58%). Interobserver and intraobserver differences between mean calcium scores were not significant, and consequently, lower interobserver and intraobserver variabilities (narrow interquartile ranges of 0-5%) were observed for all scores. Despite significant interscan differences in calcium scores, the percentile ranking assigned to the two scans differed in only 13% of patients. Interobserver differences resulted in a change in the percentile ranking in 7-9% of patients, whereas intraobserver differences caused a change in only 5% of patients.
Conclusion: The accuracy of sequential MDCT for coronary calcium quantification is sufficient in most cases for stratification of patient risk.