Introduction

Despite major progress in immunosuppressive therapy, cardiac allograft vasculopathy (CAV) remains the main cause of morbidity and mortality in the late phase after cardiac transplant [1]. CAV is characterized by concentric myointimal thickening of intramural and epicardial arteries, which leads to vascular stenosis and occlusion, resulting in graft failure and graft loss [2]. As a result of the denervation of the transplanted heart, CAV is initially a clinically silent disease that presents in a later stage of disease with heart failure, arrhythmias and sudden cardiac death. Therefore, early detection of CAV is of paramount importance for patients, as changes in drug regimen may avoid disease progression [3, 4].

Screening of CAV is traditionally performed by invasive coronary angiography (ICA) repeated on a regular basis [5]. However, ICA does not allow evaluation of the vessel wall; therefore, the early stages of CAV may be missed if no additional intravascular ultrasound is performed [6].

During recent years, CCTA has emerged as a promising tool for the assessment of CAV [7, 8]. Contrary to ICA, CCTA is non-invasive and able to assess both luminal and vessel wall changes [9]. However, there are concerns about the radiation dose associated with CCTA in heart transplant recipients owing to the need for repetitive examination, which results in a high cumulative radiation dose [10]. With the introduction of the second generation of dual-source CT technology, enormous reductions of radiation dose have been realized [11]. However, dose reduction in CCTA for the evaluation of CAV is challenging because, as a result of the denervation of the graft, patients usually present with elevated heart rates that are not sensitive to beta-blockers. Therefore the aim of this study was to prospectively evaluate image quality (IQ) and radiation dose using three different CT imaging protocols for dual-source CCTA in heart transplant recipients.

Methods

The study was approved by the local institutional review board. Power calculation revealed a sample size of 50 per group in order to detect a mean difference in overall IQ of 1 per patient with a power of 90 %. All patients gave written, informed consent prior to study enrolment. Patients were randomized using a computer-generated list (www.randomizer.org).

Patients

A total of 150 consecutive patients (121 male/29 female) after cardiac transplant were included in this prospective randomized trial. Patients with known allergic reaction to iodinated contrast agents, hyperthyroidism and elevated serum creatinine (>1.5 mg/dl) were considered ineligible.

CT protocols

All examinations were performed using a 128-row, dual-source, multi-slice CT scanner (Somatom Definition Flash, Siemens Medical Systems, Germany), with 2 × 128 × 0.6 mm collimation and a 280 ms gantry rotation. Patients were randomized to one of the following three study protocols:

  1. 1.

    Retrospective gated spiral mode (rECG): The tube voltage was set to 120 kVp for both tubes. Tube current was set to 320 reference mAs. Adaptive tube current modulation was used, which translates into an optimal pulsing window chosen by the scanner depending on the heart rate.

  2. 2.

    Prospective triggered sequential scan (pECG): The tube voltage was set to 120 kVp for both tubes. Tube current was set to 320 reference mAs. The main padding window was set at 30–70 % of the RR ECG cycle.

  3. 3.

    Prospective triggered sequential scan with a narrow systolic window (pECGsys): The main pulsing window was set to 35–45 % of the RR cycle. Automated tube voltage settings, calculated automatically and based on the acquired topogram (CAREkv), were switched on. Reference mAs values were chosen by the scanner, depending on the kVp settings.

For the completion of the prospective triggered sequential scans 3–4 slabs (each 64 × 0.6 mm) were required.

Contrast agent injection protocols

Contrast agent injections were performed using a programmable power injector (CT Expres, Swiss Medical Care, Switzerland). A non-ionic iodinated contrast medium was used at a concentration of 400 mg iodine/mL (Iomeron 400, Bracco, Austria). The flow was set to 5 mL/s and the amount of the contrast medium depended on the body weight of the patients (70–110 mL). A bolus tracking technique was applied, with the reference image at the level of the ascending aorta (threshold 150 HU; delay 5 s).

Image reconstruction

Images were reconstructed at 0.6 mm slice thicknesses on a Syngo® workstation (Siemens Medical System, Germany) using filtered back-projection and a medium soft kernel (B26F). An increment of 10 % RR was used for image reconstruction. Curved planar reformats for all three main epicardial vessels, as well as thin maximum intensity projections (MIP), were additionally reconstructed. MIP images were reconstructed in the systolic and diastolic phase in rECG and pECG, whereas only the systolic phase was provided in pECGsys.

Image quality rating

For IQ analysis, the coronary artery tree was divided into 16 segments. Fifteen segments were classified according to the American Heart Association (AHA) model, and the intermediate branch artery was assigned to segment 16 [12]. For further analysis, coronary artery segments were divided into large (AHA segments 1–3, 5–7, 11, 13) and small segments (4, 8–10, 12, 14–16). Small vessels with a diameter below 1.5 mm were not excluded from the analysis. Two experienced readers performed the IQ reading separately, using a four-point scale:

  • Grade 1 (excellent): excellent IQ enabling sufficient differentiation of even small structures; very good vessel wall visualization.

  • Grade 2 (good quality): good IQ providing all details necessary for an adequate diagnosis; good vessel wall visualization.

  • Grade 3 (poor-diagnostic): only patency or occlusion of the vascular segment can be assessed; details of vessel wall pathologies cannot be distinguished.

  • Grade 4 (non-diagnostic): vascular segments cannot be assessed; details of vessel wall pathologies are not provided.

Occluded vessel segments and segments that were not visible owing to anatomical variance were graded zero.

For IQ group comparison, the means of the per segment ratings from both readers were calculated. A higher mean per segment score translates into a reduced IQ.

Radiation dose calculation

For the estimation of the radiation dose, the dose length product (DLP) was calculated by the CT scanner by multiplication of the computed tomography dose index (CTDI) by the imaging range. Effective dose (ED) was subsequently calculated by using a standard conversion factor of 0.014 mSv mGy−1 cm−1 [13].

Statistical analysis

Statistical analyses were performed using the Statistical Package for the Social Sciences, version 20.0 (SPSS, Chicago, Illinois). Descriptive statistics were used for nominal data. An intraclass correlation coefficient (ICC) was calculated to test for agreement between the two raters. For each subject, three additional variables were defined to serve as a measure of IQ: means were calculated for large, small and for all segments on the basis of the mean ratings from both raters. Segment ratings that were rated zero (i.e. occluded or not visible) were excluded to avoid any bias in the group comparison. An analysis of variance was conducted to test for differences between the three different subgroups. The alpha level for all tests was set at p = 0.05, and correction for multiple comparisons was performed using least significant difference (LSD).

Results

Patient population

Mean age, mean BMI and mean heart rate showed an equal distribution, without significant differences among the study groups (Table 1).

Table 1 Patient and scanning characteristics: nominal data is presented as mean ± SD (range)
Full size table

Group comparison of image quality

In all patients, 339 (14.1 %) of the coronary artery segments were rated zero owing to anatomic variance or occlusion and were excluded from the further analysis (rECG: n = 110 [13.8 %]; pECG: n = 111 [13.9 %]; pECGsys: n = 118 [14.8 %]).

For the calculation of the ICC, all 150 patient ratings were included. ICC revealed excellent agreement between the two raters (0.905, CI [0.868; 0.931]).

Descriptive statistics from IQ reading are provided in Table 2. A non-diagnostic image (grade 4) quality rating was assigned, overall, to 0.3 % (n = 7) of the segments (rECG: 0, n = 0; pECG: 0.7 %, n = 5; pECGsys: 0.3 %, n = 2).

Table 2 Descriptive statistics from group comparison of image quality reading
Full size table

Overall mean IQ per segment rating was 1.45 ± 0.36, IQ rating of the large epicardial segments reached a mean of 1.33 ± 0.35, and both showed no significant differences between the study groups. Analysis of the small segments delivered an overall mean per segment score of 1.62 ± 0.43. Post hoc tests of the small segments resulted in a significant difference between the rECG and the pECGsys group (1.50 ± 0.40 vs. 1.70 ± 0.43, p = 0.02). Mean per segment score of the pECG group was 1.65 ± 0.48, with no statistical significance compared to the other groups (Fig. 1; Table 3).

Fig. 1
figure 1

Results from group comparison of coronary artery segment image quality rating. Box-and-Whisker plots show the median and 50 % of cases within the boxes as well as all data excluding outliers between the end-points of the whiskers. n.s. not significant

Full size image
Table 3 Results from post hoc tests of mean image quality per segment
Full size table

Radiation dose

Group comparison of the radiation dose values showed significant differences between pECG vs. pECGsys (p < 0.001) and rECG vs. pECGsys (p < 0.001) (Fig. 2; Table 4). In pECGsys, one patient was examined at 80 kV, 22 patients were examined at 100 kV, 25 patients were examined at 120 kV and one patient was examined at 140 kV.

Fig. 2
figure 2

Comparison of effective dose between three study protocols. Box-and-Whisker plots show the median and 50 % of cases within the boxes as well as all data excluding outliers between the end-points of the whiskers. n.s. not significant, mSv millisievert

Full size image
Table 4 Results from the group comparison of radiation dose values
Full size table

Discussion

The results of our study suggest that the radiation dose of CCTA in heart transplant recipients can be significantly reduced by using an ECG-gated sequential scan technique, with scanning in the systolic phase only, and automated tube voltage selection. Using prospective imaging and placing a narrow padding window into the systolic phase only, we maintained the overall diagnostic IQ in cardiac allograft recipients presenting with elevated heart rates. However, IQ in our study was found to be slightly lower compared to retrospective gating, in particular with regard to the small coronary artery segments; however, the IQ remains at an acceptable level.

In this study of 150 patients, we could further show that, despite higher mean heart rates, overall IQ and the assessability of coronary segments from a dual-source CCTA exam were excellent, with an excellent inter-rater correlation (Fig. 3).

Fig. 3
figure 3

CCTA in patients with cardiac transplant using three different scan protocols. a–c Curved planar reformation of the left anterior descending artery in patients with no evidence of allograft vasculopathy. d, f Curved planar reformation of the right coronary artery in patients with allograft vasculopathy. Similar image quality is evident in the pECGsys group with considerably reduced radiation dose values in c, f. rECG retrospective gating, pECG prospective gating, pECGsys systolic prospective gating + automated tube voltage selection, mSv millisievert

Full size image

Different imaging techniques may be used for CCTA [11, 14]. The retrospective gating technique with dose-modulation algorithms, used in rECG, is defined by low-pitch scanning with full tube current throughout a predefined window of the cardiac cycle, while the rest of the cardiac cycle is scanned with reduced tube current. In our study, rECG showed the lowest mean per segment ratings, representing the best IQ. However, there were no significant differences with regard to overall IQ and IQ of the main epicardial coronary artery segments compared to the other groups. At the same time, the highest radiation dose values were present in the rECG group.

The prospective ECG-gated sequence technique used in pECG is defined by applying the full tube current only at predefined phases of the cardiac cycle. The magnitude of dose-saving potential depends on how narrow the padding window is set. Commonly, for elevated heart rates, a wider padding window has been suggested to ensure the availability of multiple phases for image reconstruction. Therefore, no significant dose savings when compared to rECG were evident in our study. With regard to IQ, no differences were found when pECG was compared to the other study groups. In conclusion, prospective gating with a wide padding window does not offer any benefits to cardiac allograft recipients when compared to retrospective gating technique.

The prospective ECG-gated sequence technique with scanning in the late systolic phase only has recently been reported as feasible in patients with elevated heart rates and atrial fibrillation, while reducing radiation dose at the same time [15, 16]. Using this method, scanning is only performed in a narrow systolic phase from 35 to 45 % of the cardiac cycle. Bastarrika et al. proved that image reconstructions obtained from the systolic phase in CCTA of heart transplant recipients showed excellent IQ, concluding that this approach seems to be suitable for the evaluation of CAV [17]. In our study, the application of a systolic pulsing window showed no significant differences with regard to overall IQ or for IQ of the large epicardial arteries. In contrast to previous studies, we evaluated IQ of the large and the small coronary artery segments separately, as CAV may also involve small segments at an early stage [17]. Analysis of the small segments revealed a significant difference between rECG and pECGsys, with lower IQ in the pECGsys group. Although we obtained a statistically significant difference, it remains questionable whether this result is also clinically relevant and implies that scanning in the systolic phase only should not be used for CAV assessment. A mean per segment score of 1.7 represents good diagnostic IQ, still suitable for the evaluation of CAV.

In addition to tailoring the CT protocol to the heart rate, further strategies to achieve a dose reduction include automated, attenuation-based tube voltage selection [18]. While using this technique, kVp values of 80, 100, 120 and 140 are chosen by the scanner system, based on the pre-scan topogram [19]. This offers the possibility of significant radiation dose reduction, without the need for manual intervention [20]. In our study, the combined application of automated tube voltage selection and prospective systolic imaging resulted in a significant dose reduction of 54 %, compared to retrospective gating, and 50 %, as compared to prospective gating, using a wide padding window.

Among the diverse imaging modalities used for the follow-up of heart transplant recipients, ICA is considered the gold standard [5]. However, its invasive nature, and the inability to assess the vessel wall, has led to further development of alternative non-invasive techniques [21]. During the last decade, CCTA has emerged as a robust imaging modality for the evaluation of CAV, with sensitivity values ranging from 71 to 100 % and high negative predictive values of 93–100 % when compared to ICA [9]. A recent meta-analysis by Wever-Pinzon et al., including mainly studies using single-source and first-generation dual-source CT systems, strongly suggests CCTA for the screening of CAV, in order to avoid invasive procedures [7]. However, most of the papers included in this meta-analysis did not use any specific CT protocols tailored to imaging CAV at a lowered radiation dose, and some did not even report on radiation dose at all. On the other hand, radiation dose has received more attention during recent years, as there are major concerns about the increase in cumulative radiation dose from medical imaging [10, 22]. Cardiac transplant patients have a 3.5 times greater exposure to radiation and receive a mean cumulative radiation dose of 84 mSv during a 10-year follow-up [23]. Therefore, dose reduction in these patients is of even higher relevance. Using 64-slice scanners, high average dose values from 8 to 18 mSv have been reported in CCTA of heart transplant recipients, and, despite the technical progress, patients with arrhythmia and elevated heart rate still represent a so-called difficult-to-image population in CCTA [24].

Within the last decade, tremendous progress has been made in terms of radiation dose reduction in CCTA. Radiation dose exposure in CCTA has dropped from 15–20 mSv down to 1 mSv [25]. However, most of the studies reporting on low dose in CCTA included non-obese patients with low and regular heart rates, but hardly any patients with heart rate values above 70 bpm [26]. Although not yet evaluated in comparison to the gold standard ICA, CCTA, using recent scanner techniques, may overcome these limitations and may become a reliable non-invasive test for CAV even at lowered dose, as shown by our study results.

We acknowledge some study limitations. A comparison to the gold standard ICA is lacking, as not all patients underwent consecutive ICA within a reasonable time frame [5]. Second, patients examined a long time after transplantation might have reacted to beta-blockers because of cardiac reinnervation, which might have had a benefit on radiation dose and IQ. When looking at the CT technique, two dose-saving algorithms were combined in pECGsys and, from our data, it is not clearly distinguishable how much each of these factors contributed to the overall dose reduction. However, this approach represents the future direction of modern CT, in order to achieve a maximum dose reduction [11, 27]. Furthermore, the positive trend of a superiority of rECG over pECGsys regarding overall IQ would have probably been reached if more patients had been included. Nevertheless, although this significance was not met in the present study, this would not necessarily affect the clinical recommendations.

In conclusion, automated tube voltage selections, combined with prospective systolic scanning, enable a significant dose reduction of approximately 50 %, compared to retrospective scanning, with comparable overall diagnostic IQ and lowered, but acceptable, IQ in the small coronary artery segments. In the context of recently published papers, this underlines the role of CCTA in the evaluation of CAV [7, 8]. With further reduction of radiation dose values using the next generation of CT scanner systems, patient-tailored scan protocols and iterative reconstruction algorithms, CCTA could take the next step and become a useful, non-invasive screening method for CAV, replacing ICA [28].