Three-dimensional rotational angiography in children with an aortic coarctation

This retrospective study analysed CA and 3DRA-guided cardiac catheterisations that were performed in CoA patients at the Wilhelmina Children’s Hospital, University Medical Center Utrecht between January 2003 and December 2015. All children who underwent cardiac catheterisation for aortic arch interventions were included. For designing the 3D-imaging protocol all children who underwent a 3DRA run during a diagnostic or interventional catheterisation were included. The institutional medical ethics committee provided a waiver for this study.

Demographic, clinical and catheterisation data and technical settings of the 3DRA were retrospectively obtained from patient files, cardiac catheterisation database (Filemaker Pro 11), and image databases (Xcelera 4.1, OsiriX MD 2.6 and Siemens syngoDynaCT).

Fluoroscopy films were included in the analysis, but still frames were excluded. Diagnostic angiographies were defined as all images with contrast displaying relevant cardiac compartments and vessels. Interventional angiographies were defined as all images with contrast displaying balloons or stents and post-interventional control images. In diagnostic angiographies, visibility of the left ventricle, aorta, coronary arteries and ostium of the left subclavian artery was scored on a 3-point scale. Good image quality was defined as providing a clear delineation of cardiovascular structures, moderate image quality as providing an estimated delineation and poor image quality as providing no delineation. Image quality was determined by the primary researcher and in case of doubt by paediatric cardiologists (GK and/or JB).

Therapeutic success was defined as a post-interventional systolic gradient across the CoA under general anaesthesia of <20 mm Hg [12]. Complications were collected in accordance with guidelines [13]. Radiation is reported as dose area product (DAP), which is defined as the product of radiation dose and exposed patient surface [11]. Two paediatric cardiologists (GK and JB) determined the additional value of 3D-images compared with CA images according to a 5-point scale, ranging from essential to misleading [6, 7].

To analyse radiation, fluoroscopy time and number of angiographies performed, three groups were created: patients with sole diagnostic CAs, both diagnostic CAs and 3DRAs and only diagnostic 3DRAs. Diagnostic quality was analysed in the full population, as this is not influenced by the subsequent type of intervention. Further analyses were performed in patients undergoing either balloon angioplasty (BA) or single stent implantation to enhance comparability. Patients receiving multiple stent implantations were excluded.

3DRA runs of perfect diagnostic quality without artefacts, with the exception of stent artefacts, were used to design the 3D-imaging protocol. Patients with incomplete data sets were excluded.

In total, 134 patients underwent 183 catheterisations. Sixty-seven patients underwent BAs, 61 CAs (80.3 %) and 15 3DRAs (19.7 %). Sixty-four patients received stents, 43 CAs (61.4 %) and 27 3DRAs (38.6 %). Median age was 0.52 years in the BA group and 11.19 years in the stent group (p < 0.001). In the BA group, maximum Doppler velocity in the descending aorta was significantly lower in the 3DRA group (p = 0.020). In the stent group, the 3DRA group were significantly older (p = 0.045) (Table 1).

3DRA was superior in displaying the left ventricle, ascending aorta, aortic arch and coronary arteries. 3DRA displayed the descending aorta and ostium of the left subclavian artery equally as well as CA (Fig. 1). Artefacts were significantly more frequently encountered in 3DRAs (25.8 %) than in CAs (5.1 %) (p < 0.001). In the 3DRA group, artefacts were mostly caused by catheter movements (N = 7 (43.8 %)) due to placement of the angiographic catheter in the ascending aorta (N = 5) or aortic arch (N = 1).

In the BA group, the 3DRA group had a significantly higher success rate than the CA group: 100.0 % versus 68.9 % respectively (p = 0.016). All stent interventions were successful. In the BA group, there was a trend towards lower median residual gradients in the 3DRA than in the CA group (9.00 mm Hg versus 13.00 mm Hg, p = 0.087). In the stent group, systolic and diastolic blood pressure were significantly higher in the 3DRA than in the CA group (p = 0.002 and p = 0.044 respectively) (Table 2).

In the BA group, procedural complications occurred in 8 CA catheterisations (13.1 %) versus 4 3DRA catheterisations (26.7 %) (p = 0.238) and interventional complications in 18 CA catheterisations (29.5 %) versus 3 3DRA catheterisations (20.0 %) (p = 0.538). In the stent group, procedural complications occurred in 15 CA catheterisations (34.9 %) versus 7 3DRA catheterisations (25.9 %) (p = 0.432) and interventional complications in 1 CA catheterisation (2.3 %) versus 0 3DRA catheterisations (p = 1.000). Additional Table 4 displays the different types of complications.

3DRA was associated with fewer additional interventional CAs in both the BA and stent group. This difference was significant for the A‑plane angiographies in the stent group (p = 0.021). In the BA group, the median total number of additional CAs was 6.00 (4.00–8.00) in the CA, and 2.00 (0.50–2.00) in the 3DRA group (p = 0.001). In the stent group, this was 16.00 (12.00–18.00) in the CA, and 8.00 (4.00–9.50) in the 3DRA group (p < 0.001) (Additional Table 5).

In the BA group, all DAPs were lower in the 3DRA than in the CA group. In the stent group, DAP due to additional angiographies was significantly lower in the 3DRA than in the CA group (p = 0.042) (Table 3).

The additional value of 62 3DRAs in CoA interventions was evaluated. The scores were as follows: 10 3DRAs were essential (16.1 %), 44 very useful (71.0 %), 6 useful (9.7 %), 2 not useful (3.2 %) and 0 misleading (0.0 %). Not useful 3DRAs were of poor diagnostic quality due to suboptimal settings of the 3DRA. The greatest benefit of 3DRA was the ability to fully understand aortic arch and CoA morphology (N = 60 (96.8 %)) (Fig. 2). Other benefits were the possibility of displaying other cardiovascular structures (N = 17 (27.4 %)) and vascular anomalies (N = 17 (27.4 %)) in the same run (Fig. 3). In univentricular heart patients, 3DRA was useful in displaying vessel-vessel or vessel-bronchi interactions (N = 5 (8.1 %)) (Fig. 3).

The total number of 3DRA-guided CoA catheterisations increased from 12 in 2012 to 23 in 2015. Conversely, the median total DAP strongly decreased from 87.06 μGym²/kg to 23.20 μGym²/kg in the BA group and from 90.15 μGym²/kg to 43.01 μGym²/kg in the stent group.

77 patients received 126 3DRA runs. Two 3DRA runs were performed for pre- and post-procedural imaging (N = 28), technical difficulties (N = 4) and requirement of re-intervention (N = 1). Seventy-one 3DRAs were suitable for designing the imaging protocol (Fig. 4).

Image quality of the left ventricle, ascending aorta, aortic arch and coronary arteries is significantly higher in 3DRA than in CA. These structures are important for good device positioning. Studies with varying areas of interest report diagnostic quality rates of 3DRA from 71 to 94 % [8, 9, 14]. The diagnostic quality in this study is even higher at 98.4 % good image quality of the aortic arch and subclavian artery. This could be due to consistent use of rapid pacing and X‑ray delay to ensure optimal contrast filling [8, 9, 14]. Image quality was similar in the BA and stent groups despite the significant age difference. However, 3DRA reconstructions contain more artefacts than CA. Some artefacts can be resolved by changing the technical settings of 3DRA and post-processing. In all other, the dynamic rotational angiogram is almost always diagnostic.

Therapeutic success was high in both the BA and stent groups. 3DRA-guided BA resulted in lower residual gradients than previously reported (9.00 versus 12.5 mm Hg) [4, 15] and was superior to CA in our study. This may be the result of superior diagnostic imaging leading to improved interventional therapy or more aggressive treatment strategies. Residual gradients after stenting were low and comparable to other studies in both CA and 3DRA group [4, 16]. To date, no study we are aware of has performed similar comparisons.

Interventional complication rates are comparable to those described elsewhere [4, 13]. Only one stent dislocation occurred in the CA group. One can speculate that additional 3D-imaging helps to optimise stent sizing and positioning, thereby reducing the risk of stent dislocation.

In five years, a threefold decrease in radiation exposure was accomplished during 3DRA procedures. There were several reasons for this: first, technical settings were altered and frame rate was lowered from 60 to 30 frames/s without disadvantageous effect on image quality. Second, 3D-images are of superior diagnostic quality which reduces the need for additional CAs. Third, during the learning curve, paediatric cardiologists initially performed both CA and 3DRA, but with increasing knowledge and confidence, this was abandoned.

The radiation dose in the 3DRA group was not significantly higher than that in the CA group (p = 0.275 for BAs; p = 0.090 for stents). This contrasts with the study from Manica et al., which reported a significantly higher DAP of a single 3DRA run compared with three CAs in patients under 45 kg [8]. Total DAPs were comparable to other studies, which reported median DAPs for various congenital heart disease interventions of 3605 μGym² [11] and 1429.6 μGym² for CoA stenting [7]. More importantly, we see a clear decrease in DAP, currently resulting in very low total DAPs compared with benchmarks (median) of 2000 μGym² in children between 1–4 years old and 9600 μGym² in children between 10–15 years old [17]. The remaining radiation difference between BA and stenting is due to a difference in complexity of the procedure and age and therefore size of the patients.

3DRA was of additional value in 96.8 % of the interventions, which is similar to other cohorts. However, we found more essential 3DRAs [6, 7]. This can be caused by obtaining more information from 3DRA, as experience with this technique increases.

The greatest advantage of 3DRA is the ability to fully understand aortic arch and CoA morphology. 3DRA is able to visualise the aorta from an infinite number of angles, some of which are very important, such as the cranioposterior angle, which are not provided by CA [6‐9]. Furthermore, other vascular and extravascular (e. g. airway) structures can be visualised simultaneously with a high spatial resolution [8]. Moreover, iPilots are of great additional value for guidewire and stent positioning, thereby reducing the need to perform multiple CAs to guide the procedures [6, 7, 9].

We used 3DRA runs of perfect diagnostic quality to design a 3DRA imaging protocol for CoAs. This is the first practical guide ever on optimisation of 3DRA and may help others to start up 3DRA catheterisation laboratories with faster learning curves.

This study is limited by its retrospective nature. In addition, our learning curve influenced some median values, such as radiation doses. Finally, scoring of image quality and additional value of 3DRA cannot be purely objective. However, we tried to limit the subjectivity by using predefined scores.