OBJECTIVE: Dynamic image analysis of carotid plaques has demonstrated that during systole and early diastole all plaque components move in the same direction (concordant motion) in some plaques, while in others, different parts of the plaque move in different directions (discordant motion). The aim of our study was (a) to determine the prevalence of discordant motion in symptomatic and asymptomatic plaques, (b) to develop a measurement of the severity of discordant motion and (c) to determine the correlation between the severity of discordant motion and prevalence of symptoms.
METHODS: 200 patients with 204 plaques producing 50-99% stenosis (112 asymptomatic and 92 symptomatic) had video recordings of the plaque motion during 10 cardiac cycles. Video tracking was performed using Farneback's method which relies on frame comparisons which in our study were made at 0.1 second intervals. The maximum angular spread (MAS) of motion vectors at 10-pixel intervals in the plaque area was measured in degrees. Plaques were classified as concordant (MAS < 700), moderately discordant (MAS 70-1200) and discordant (MAS > 1200). RESULTS: Motion was discordant in 89.1% of symptomatic plaques and only in 17.9% of asymptomatic plaques (p < 0.001). The prevalence of symptoms increased with increasing MAS. For MAS > 1200 the hazard ratio for presence of symptoms was 47.7 (95% CI 18.1 to 125.6) compared with the rest of the plaques after adjustment for degree of stenosis and mean pixel motion. The area under the ROC curve for prediction of the presence of symptoms by MAS was 0.876 (95% CI 0.823 to 0.929). The use of the median value of MAS (1200) as a cut-off point, could classify 86% of plaques correctly (sensitivity 81.4%; specificity 91.2%; positive predictive value 90.2% and negative predictive value 83.0%).
CONCLUSIONS: The value of MAS in identifying asymptomatic plaques at increased risk of developing symptoms and particularly stroke should be tested in prospective studies.