Defining coronary artery calcium concordance and repeatability - Implications for development and change: The Dallas Heart Study.
J Cardiovasc Comput Tomogr. 2017 Sep - Oct;11(5):347-353. doi: 10.1016/j.jcct.2017.06.004. Epub 2017 Jul 3. Paixao ARM1, Neeland IJ2, Ayers CR2, Xing F3, Berry JD2, de Lemos JA2, Abbara S2, Peshock RM2, Khera A4.
BACKGROUND: Development and change of coronary artery calcium (CAC) are associated with coronary heart disease. Interpretation of serial CAC measurements will require better understanding of changes in CAC beyond the variability in the test itself.
METHODS: Dallas Heart Study participants (2888) with duplicate CAC scans obtained minutes apart were analyzed to determine interscan concordance and 95% confidence bounds (ie: repeatability limits) for each discrete CAC value. These data derived cutoffs were then used to define change above measurement variation and determine the frequency of CAC development and change among 1779 subjects with follow up CAC scans performed 6.9 years later.
RESULTS: Binary concordance (0 vs. >0) was 91%. The value of CAC denoting true development of CAC by exceeding the 95% confidence bounds for a single score of 0 was 2.7 Agatston units (AU). Among those with scores >0, the 95% confidence bounds for CAC change were determined by the following formulas: for CAC≤100AU: 5.6√CAC + 0.3*CAC - 3.1; for CAC>100AU: 12.4√CAC - 67.7. Using these parameters, CAC development occurred in 15.0% and CAC change occurred in 48.9%. Although 225 individuals (24.9%) had a decrease in CAC over follow up, only 1 (0.1%) crossed the lower confidence bound. Compared with prior reported definition of CAC development (ie: >0), the novel threshold of 2.7AU resulted in better measures of model performance. In contrast, for CAC change, no consistent differences in performance metrics were observed compared with previously reported definitions.
CONCLUSION: There is significant interscan variability in CAC measurement, including around scores of 0. Incorporating repeatability estimates may help discern true differences from those due to measurement variability, an approach that may enhance determination of CAC development and change.