Population
Our study group was drawn from a population of 249 consecutive oncology patients gathered in two successive time periods: from May 2008–December 2008 a prototype version of the high temporal resolution-high pitch mode with a first-generation dual-source CT system was used on 123 consecutive patients, forming group 1 (temporal resolution: 83 ms); from September 2009-April 2010 a commercial version of the high temporal resolution-high pitch mode on the second generation of dual-source CT system was used of 126 consecutive patients, forming group 2 (temporal resolution: 75 ms). Patients in group 1 and 2 (1) were all referred from the same pulmonology department for follow-up of thoracic malignancy; (2) had no concurrent abnormalities that could influence coronary artery imaging; and (3) had an upper limit of 100 kg for the patient’s body weight (b.w.) to ensure full coverage of both lungs by the tube B system (size of the tube B detector: 26 cm in group 1; 33 cm in group 2). Storage of data sets on external disks enabled their retrospective analysis with a waiver of patients’ informed consent in agreement with national regulations.
CT examination
Acquisition parameters
Apart from the rotation time (and the size of the tube B detector), the acquisition parameters were similar for patients in group 1 and 2, including (1) a craniocaudal acquisition at deep inspiration; (2) detector collimation: 2 × 64 × 0.6 mm, slice acquisition 2 × 128 × 0.6 mm by means of a z-flying focal spot; (3) the kilovoltage was set at 100 kV (reference milliamperage: 90 mAs) for patients below 80 kg b.w. and at 120 kV (reference milliamperage: 90 mAs) for patients with a body weight between 80 and 100 kg; (4) the examinations were systematically obtained with automatic adjustment of the milliamperage according to the patient’s size (Care Dose 4D; Siemens Medical Solutions, Forchheim, Germany) without administration of beta-blockers or sublingual nitrates.
Injection parameters
In both groups, the injection protocols were similar [nonionic contrast material: iohexol; Omnipaque 350; Amersham Health, Carrigtohill, Ireland; volume: 120 ml (this corresponds to the maximal volume in the injector; the volume administered was often smaller as the injection was manually interrupted whenever necessary at the end of the acquisition); concentration: 350 mg of iodine per millilitre; flow rate: 4 ml/s without saline flush]. The same automatic bolus triggering software program available on the two CT units (Care Bolus; Siemens Medical Solutions, Forchheim, Germany) was systematically applied (ROI within the ascending aorta; threshold: 150 HU).
Reconstruction parameters
Cross-sectional images were reconstructed in both groups from the carina to the heart base with a slice thickness of 0.75 mm in 0.6-mm intervals, a small field of view cropped around the heart, and a soft-tissue convolution kernel (B26f) to generate curved MPRs and MIP images of the coronary arteries (Leonardo workstation; Circulation software; Siemens Medical Solutions, Germany).
Parameters evaluated
Patients’ characteristics
For each examination, the patient’s heart rate, breath-hold duration, z-axis coverage, and dose-length product (DLP) were systematically recorded. The effective dose was derived from the product of the DLP and a conversion coefficient (k = 0.017 mSv.mGy-1 cm-1) [8]. Although different conversion factors exist and may result in higher dose estimates, this value was selected to allow comparison with recent findings in the literature.
Accessibility of coronary arteries
Image analysis was obtained by consensus between two readers with 2 years and 7 years of experience in cardiac CT at the time of initiation of this study, respectively. The images were reviewed on a workstation (MMWP, Siemens, Forchheim, Germany). The coronary arteries were categorised into 15 segments according to the American Heart Association classification [9]. For each patient, the analysis of the coronary artery tree was undertaken only if the attenuation value within all proximal coronary segments was ≥200 HU. The next step consisted of an individual analysis of each coronary segment (whatever its diameter and the potential presence of calcification), rated as accessible if the vessel lumen was completely depicted in the absence of major motion artefacts as further detailed. The degree of movement-associated artefacts was graded with a 4-point scoring system (10): score 1: no motion artefacts; score 2: mild blurring of the segment; score 3: moderate blurring without structure discontinuity; score 4: doubling or discontinuity in the course of the segment preventing evaluation. Scores 1–3 were considered compatible with a diagnostic image quality. The impact of contrast-induced artefacts from the right atrium on the right coronary artery was graded with a 3-point scoring system: score 1: no artefact; score 2: minor artefacts, not precluding analysis of the coronary segment; score 3: marked artefacts, precluding confident analysis of the coronary segment. No attempt was made to provide information on the frequency of coronary artery abnormalities detected in either group of patients.
Four levels of analysis of the coronary tree were defined: (1) analysis of proximal segments (i.e. 4 segments per patient), including segment 1 of the right coronary artery (RCA), segment 5 (left main, LM), segment 6 of the left anterior descending (LAD), and segment 11 of the left circumflex (LCX); (2) analysis of proximal and mid segments of the coronary arteries, namely segments 1–2 (RCA), 5 (left main, LM), 6–7 (LAD), and 11 and 13 (LCX), leading to the analysis of seven segments per patient; (3) analysis of ten segments per patient, including the above-mentioned seven segments and segment 3 (right coronary artery, RCA), segment 4 (right coronary artery, RCA, in case of right predominance; left circumflex artery, LCX, in case of left predominance), and segment 8 (left anterior descending artery, LAD); (4) analysis of the complete coronary arterial tree, i.e. 15 coronary segments per patient.
Statistical analysis
Statistical analysis was performed with commercially available software (SAS Institute, Cary, NC). Results were expressed as means and standard deviations for continuous variables and as frequencies and percentages for categorical variables. Comparative analyses were obtained using the chi-square test for categorical data; when not applicable because of the sample size, Fisher’s exact test was used. For numerical variables, we used the unpaired Student’s t-test. P values inferior to 0.05 were considered statistically significant.