Reproducibility of two coronary calcium quantification algorithms in patients with different degrees of calcification

To evaluate the reproducibility of coronary calcium quantification algorithms by electron beam CT (EBT) in patients with different amounts of calcified plaque using the conventional (Agatston) score and an area score and to demonstrate a potential application of these results for evaluation of follow-up scans.

In 50 consecutive patients. the conventional calcium score (CCS = Agatston score) and the area score (AS) were summed for each artery and patient. Data were analyzed in four groups according to degrees of calcification: 0 (absent-minimal): CCS 0-9, I (mild): CCS 10-99, II (moderate): CCS 100-399, III (severe): CCS > or = 400. We determined and compared the reproducibility for each algorithm within and among groups.

Median percent reproducibility improved with increasing amounts of calcified plaque for the CCS and the AS (p = 0.002 and p = 0.004, respectively). We demonstrate how these reproducibility values can be used to evaluate long-term follow-up studies. The reduction of median reproducibility per patient using the AS vs. the CCS was 32% (13 vs. 19%, respectively). On a vessel-by-vessel basis, the reduction of median reproducibility was 7% (24.3 vs. 22.6%, CCS vs. AS, p < 0.02), which was attributable to a 45% reduction in reproducibility in arteries with mild scores (46.1 vs. 25.5%, CCS vs. AS, p < 0.005).

The AS has an improved reproducibility compared with the CCS, especially in patients with small amounts of coronary calcifications which may prove clinically useful. Different reproducibility values in different degrees of calcification can be used for an individual assessment of changes in amounts of coronary calcification.