The added value of longitudinal black-blood cardiovascular magnetic resonance angiography in the cross sectional identification of carotid atherosclerotic ulceration

Abstract

Background: Carotid atherosclerotic ulceration is a significant source of stroke. This study evaluates the efficacy of adding longitudinal black-blood (BB) cardiovascular magnetic resonance (CMR) angiography to cross-sectional CMR images in the identification of carotid atherosclerotic ulceration.

Methods: Thirty-two subjects (30 males and two females with ages between 48 and 83 years) scheduled for carotid endarterectomy were imaged on a 1.5T GE Signa scanner using multisequence [3D time-of-flight, T1, proton density, T2, contrast enhanced T1], cross-sectional CMR images and longitudinal BB CMR angiography (0.625 x 0.625 mm/pixel). Two rounds of review (round 1: cross-sectional CMR images alone and round 2: cross-sectional CMR images plus longitudinal BB CMR angiography) were conducted for the presence and volume measurements of ulceration. Ulceration was defined as a distinct depression into the plaque containing blood flow signal on cross-sectional CMR and longitudinal BB CMR angiography.

Results: Of the 32 plaques examined by histology, 17 contained 21 ulcers. Using the longitudinal BB CMR angiography sequence in addition to the cross-sectional CMR images in round 2, the sensitivity improved to 80% for ulcers of at least 6 mm3 in volume by histology and 52.4% for all ulcers, compared to 30% and 23.8% in round 1, respectively. There was a slight decline in specificity from 88.2% to 82.3%, though both the positive and negative predictive values increased modestly from 71.4% to 78.6% and from 48.4% to 58.3%, respectively.

Conclusion: The addition of longitudinal BB CMR angiography to multisequence cross-sectional CMR images increases accuracy in the identification of carotid atherosclerotic ulceration.

Figure 1

Example of carotid multisequence cross-sectional CMR images and a longitudinal black-blood CMR (BB CMR) angiography. Four contrast-weighted CMR images [3-dimensional time-of-flight (TOF), T1-weighted (T1W), T2-weighted (T2W), and contrast-enhanced T1-weighted (CE-T1W)] are matched using the carotid bifurcation (dashed line) seen on the longitudinal BB CMR angiography as the landmark. Proton density weighted image is not shown. Int = internal carotid artery. Ext = external carotid artery. Common = common carotid artery. JV = jugular vein.

Figure 2

Ulcer identification using multicontrast, cross-sectional CMR images and longitudinal black-blood CMR (BB CMR) angiography. The ulcer (arrows), with a disorganized blood flow signal pattern shown on the cross-sectional CMR images, is well demarcated on the longitudinal BB CMR angiography and confirmed by corresponding histology. Proton density weighted image is not shown. Int = internal carotid artery. Ext = external carotid artery. Common = common carotid artery.

Figure 3

Example of partial volume and flow disturbance in a large penetrating ulcer. The cross-sectional CMR images alone could not verify the presence of the ulcer in the left carotid artery. The signal pattern of the ulcer resembles the signal pattern of intraplaque hemorrhage (arrows). However, the longitudinal black-blood CMR (BB CMR) angiography shows clear separation between the very narrow lumen and the penetrating ulcer due to the improvement of spatial resolution in the longitudinal direction. The flow artifact with this ulcer can be easily identified. Corresponding histology confirms the presence of the ulcer. CE-T1W is not shown. PDW = proton density weighted. Int = internal carotid artery. Ext = external carotid artery. Common = common carotid artery. JV = jugular vein.

Figure 4

Scatter plots of the ulcer volume on log scale measured by CMR and histology. Only ulcers where histology and CMR agreed in round 2 review (cross-sectional CMR images plus BB CMR angiography) were included (n = 11). Spearmen's rank correlation was calculated as r = 0.60 (p = 0.053).

Figure 5

Sensitivity of ulcer identification by CMR with regard to ulcer volume by histology. The sensitivity was calculated at each point in the Figure using only ulcers with the given volume or greater. The sensitivity was uniformly higher in round 2 review (cross-sectional CMR images plus BB CMR angiography), and in particular was notably higher for larger ulcers. Sensitivity did not appear to depend on size in round 1 review (cross-sectional CMR images alone).

Figure 6

Distributions of CMR ulcer volume measurements, grouped by whether they corresponded to true or false positives. The smaller CMR ulcer volume measurements tended to be false positives, especially in round 2 [p = 0.19 for round 1 (cross-sectional CMR images alone, and p = 0.005 for round 2 (cross-sectional CMR images plus BB CMR angiography).