Neuroradiology (2005) 47: 1–9 DOI 10.1007/s00234-004-1301-4

Alexander M. McKinney Sean O. Casey Mehmet Teksam Leandro T. Lucato Maurice Smith Charles L. Truwit Stephen Kieffer

Received: 30 December 2003 Accepted: 4 September 2004 Published online: 14 January 2005 Ó Springer-Verlag 2005

Presented at the 41st Annual Meeting of the American Society of Neuroradiology, Washington D.C., 2003. Sean Casey, MD and Charles Truwit, MD are members of the Medical Advisory Board of Vital Images (Plymouth, Minnesota), the company that developed the Vitrea 2 software. A. M. McKinney (&) Æ S. O. Casey M. Teksam Æ C. L. Truwit Æ S. Kieffer Departments of Radiology, University of Minnesota Medical School, and Hennepin County Medical Center, Minneapolis, MN 55455, USA E-mail: [email protected] Tel.: +1-612-2734092 Fax: +1-612-6248495 L. T. Lucato Clinics Hospital of the University of Sao Paulo, Sao Paulo, Brazil M. Smith Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD 21218, USA

DIAGNOSTIC NEURORADIOLOGY

Carotid bifurcation calcium and correlation with percent stenosis of the internal carotid artery on CT angiography

Abstract The aim of this paper was to determine the correlation between calcium burden (expressed as a volume) and extent of stenosis of the origin of the internal carotid artery (ICA) by CT angiography (CTA). Previous studies have shown that calcification in the coronary arteries correlates with significant vessel stenosis, and severe calcification (measured by CT) in the carotid siphon correlates with significant (greater than 50% stenosis) as determined angiographically. Sixty-one patients (age range 50–85 years) underwent CT of the neck with intravenous administration of iodinated contrast for a variety of conditions. Images were obtained with a helical multidetector array CT scanner and reviewed on a three-dimensional workstation. A single observer manipulated window and level to segment calcified plaque from vascular enhancement in order to quantify vascular calcium volume (cc) in the region of the bifurcation of the common carotid artery/ICA origin, and to measure the extent of ICA stenosis near the origin. A total of 117 common carotid artery bifurcations were reviewed. A ‘‘significant’’ stenosis was defined arbitrarily as >40% (to detect lesions before they become hemodynamically significant) of luminal diameter on CTA using NASCETlike criteria. All ‘‘significant’’ stenoses (21 out of 117 carotid bifurcations) had measurable

calcium. We found a relatively strong correlation between percent stenosis and the calcium volume (Pearson’s r = 0.65, P 40%) was seen in 16 patients (for a total of 21 carotids), where five patients had bilateral ‘‘significant’’ stenoses. For the ‘‘significant’’ stenoses, the mean calcium burden was 0.215 cc (SD 0.204). All ‘‘significant’’ stenoses demonstrated a measurable calcium burden, although one patient with prior carotid endarterectomy had a 99% stenosis and only mild calcium burden (0.04 cc). This was one of two cases of 99% stenosis in this study, both of which had previously undergone prior carotid endarterectomy on the opposite side. Three carotids with greater than 70% stenosis were detected in three different patients. Using calcium volumes of 0.01, 0.03, 0.06, 0.09 and 0.12 cc as thresholds for evaluation for ‘‘significant’’ stenoses, the sensitivity, specificity, positive predictive value, and negative predictive value were calculated (Table 1). A ROC curve was also generated using the sensitivity and specificity (Fig. 2). The sensitivity was 100% for both sides when using a minimum threshold of 0.01 cc, but the specificity was only 56%. The specificity was 93% for both sides when using 0.12 cc as a threshold, but the sensitivity was only 56%. Thresholds of 0.06 cc (sensitivity 88%, specificity 87%) and 0.03 cc (sensitivity 94%, specificity 76%) generated the best combination of sensitivity and specificity.

Table 1 Sensitivities and specificities at the five calcium volume thresholds that were used to evaluate for a ‘‘significant’’ stenosis. PPV positive predictive value, NPV negative predictive value

Sensitivity (%) Specificity (%) PPV (%) NPV (%)

0.01 cc

0.03 cc

0.06 cc

0.09 cc

0.12 cc

100.0 55.6 44.4 100.0

93.8 75.6 57.7 97.1

87.5 86.7 70.0 95.1

75.0 91.1 75.0 91.1

56.3 93.3 75.0 85.7

A stenosis of 40% or greater was selected as positive for ‘‘significant’’ stenosis. Thresholds of 0.03 cc and 0.06 cc have the highest combination of sensitivity and specificity. Statistically, the square roots of these five thresholds provide the same sensitivity, specificity, PPV and NPV so a duplicate table was not included for purposes of brevity

Fig. 2 The ROC curve is composed of five thresholds: 0.01, 0.03, 0.06, 0.09, and 0.12 cc. The ROC curve illustrates that the calcium volume thresholds of 0.03 and 0.06 cc are relatively sensitive and specific for detecting a >40% ICA stenosis

Using the data from all 117 carotid bifurcations, we computed the correlation coefficient (r) between calcium volume and percent stenosis. This correlation was moderate to strong with r = 0.65 for all carotids together (r = 0.72 for right side only, 0.62 for left only, P