Letter to the Editor
Quantification of Abdominal Aortic Calcification on CT
To the Editor:
Calcification of the coronary arteries is associated with cardiovascular end-point, such as myocardial infarction, stroke, and vascular death.1 Techniques for quantifying vascular calcification in patients, particularly in arteries outside the chest, are not routinely available.2 In this prospective study, we examined whether CT angiography (CTA) could be used to reproducibly measure severity of infrarenal abdominal aortic calcification. The study involved 50 consecutive patients undergoing aortic CTA for investigation of peripheral vascular disease or aortic aneurysm in 3 stages.
For the initial 14 patients, sequential plain and contrast-enhanced CT were obtained and then analyzed on a workstation using 5 different thresholds to define aortic calcification. The calcification volumes were assessed using a number of nonparametric statistical tests to compare clinical measurements to define the most appropriate threshold by which ca lcification should be measured on CTA by comparison with CT.
For the next 24 patients, CTA alone was obtained and analyzed with the threshold defined as optimal from part 1 of the study. The calcification volume measurements were performed by one observer using three different workstation protocols: (1) manual threshold setting without image magnification, (2) manual threshold setting with twice image magnification, and (3) automatic threshold setting with twice image magnification.
To assess the most appropriate protocol, the readings were repeated on a second occasion 24 hours apart by the same observer, and the intraobserver reproducibility was assessed by nonparametric statistical tests.
Having defined the most appropriate thresholds and workstation settings to analyze the calcification volume on CTA, the interobserver reproducibility of calcification volume measurements was analyzed on the next 12 consecutive patients undergoing CTA and compared with those for CT (using the first cohort of patients).
For details of the calcification measurement technique, please see supplemental methods and Figure I, available online at http://atvb.ahajournals.org.
Thirty-six patients presented with intermittent claudication, 14 with aortic aneurysm, 18 were female, and 12 had diabetes mellitus.
Assessment of the Most Appropriate Radiographic Threshold to Assess Aortic Calcification on CTA
Comparison of calcification volumes measured using all 5 thresholds demonstrated that the best correlation and lowest coefficient of variation (CV) between plain CT and CTA was achieved when the threshold Center Hounsfield Unit level (CH) 1400 and a Window width Hounsfield Unit (WH) 2000 were used: CV 6.7%; correlation coefficient r, 0.999; P=0.01, mean difference −53 (95% CI, 5 to −111), limit of agreement 108 (8 to 208) to −215 (−115 to −315; Table I, available online at http://atvb.ahajournals.org). Visual assessment of images at these different thresholds confirmed that of CH 1400 and WH 2000 provided good assessment of aortic wall calcification (Figure, A through F). A total CH of 1160 and WH of 2048 was noted to be a good threshold to assess aortic volume and was used for this later in the study (Figure, A through F).
Assessment of the Most Appropriate Workstation Setting to Employ By Comparison of Intraobserver Reproducibility
The Table outlines the reproducibility of readings made 24 hours apart in 24 patients by the same observer using the 3 different workstation protocols and thresholds defined from part 1 of the study. The results clearly show that the reproducibility was best when thresholds were set using the automated function and images were magnified before calcification volume measurement. Using this protocol and the threshold identified in part 1 of the study for aortic volume, good reproducibility for aortic volume measurement was also found (Table). The composite outcome of percentage calcification was found to be less reproducible (Table).
Comparison of Reproducibility of Calcification Volume Measurements by CT and CTA
The intraobserver and interobserver reproducibility of calcification volume measured with plain CT using automatic threshold setting and image magnification was good with CV <5% (Table). The reproducibility of calcification volumes measured on CTA was equally good with CV &1% (Table).
The main finding of this study is that using appropriate thresholds, CTA can be used to quantify infrarenal abdominal aortic calcification in a comparable way to plain CT. The reproducibility of aortic calcification quantification was best using automated workstation thresholds with image magnification and was comparable with that achieved with plain CT (Table). The reproducibility of calcification volume was found to be better than that for percentage calcification, likely because this reading represents a composite of 2 other readings (calcification and aortic volume) and therefore is subject to inaccuracies from both. In addition, the relatively small readings obtained for calcification percentage, because aortic volume was much greater than calcification volume, makes this reading less useful.
Compared with investigations of the coronary circulation, very few studies have assessed the ability of abdominal aortic calcification to predict cardiovascular events.3 The variation of methods used to quantify calcification in these studies (plain x-ray film, electron beam CT, and plain CT) and lack of rigorous assessment of reproducibility makes interpretation of the available data difficult.3 The use of CT to quantify abdominal aortic calcification with appropriate reproducibility assessment is likely to provide much more accurate data on the value of this finding in predicting subsequent events, which, in addition to those related to coronary calcification, may also include aortic occlusion, aneurysm formation, and outcome of interventional procedures on the aorta. The value of using CTA to carry out such quantification rather than plain CT is that both angiography and calcification data can be obtained in 1 image run, thereby reducing the dose of ionizing radiation to the patient and providing cost savings. The thresholds and workstation settings required for the calcification readings require local validation for differing software and CT scanners.
This project is supported in part by grant numbers R01 HL080010-01 from the National Institute of Health, USA and 279408 from the National Health and Medical Research Council, Australia. The authors would like to thank Philips for assistance with use of the software in this study.