Abstract
PURPOSE:
To compare standardised 2D ultrasound (US) to the novel ultrasonographic imaging techniques 3D/4D US and image fusion (combined real-time display of B mode and CT scan) for routine measurement of aortic diameter in follow-up after endovascular aortic aneurysm repair (EVAR).
METHOD AND MATERIALS:
300 measurements were performed on 20 patients after EVAR by one experienced sonographer (3rd degree of the German society of ultrasound (DEGUM)) with a high-end ultrasound machine and a convex probe (1–5 MHz). An internally standardized scanning protocol of the aortic aneurysm diameter in B mode used a so called leading-edge method. In summary, five different US methods (2D, 3D free-hand, magnetic field tracked 3D - Curefab™, 4D volume sweep, image fusion), each including contrast-enhanced ultrasound (CEUS), were used for measurement of the maximum aortic aneurysm diameter. Standardized 2D sonography was the defined reference standard for statistical analysis. CEUS was used for endoleak detection.
RESULTS:
Technical success was 100%. In augmented transverse imaging the mean aortic anteroposterior (AP) diameter was 4.0±1.3 cm for 2D US, 4.0±1.2 cm for 3D Curefab™, and 3.9±1.3 cm for 4D US and 4.0±1.2 for image fusion. The mean differences were below 1 mm (0.2–0.9 mm). Concerning estimation of aneurysm growth, agreement was found between 2D, 3D and 4D US in 19 of the 20 patients (95%). Definitive decision could always be made by image fusion. CEUS was combined with all methods and detected two out of the 20 patients (10%) with an endoleak type II. In one case, endoleak feeding arteries remained unclear with 2D CEUS but could be clearly localized by 3D CEUS and image fusion.
CONCLUSION:
Standardized 2D US allows adequate routine follow-up of maximum aortic aneurysm diameter after EVAR. Image Fusion enables a definitive statement about aneurysm growth without the need for new CT imaging by combining the postoperative CT scan with real-time B mode in a dual image display. 3D/4D CEUS and image fusion can improve endoleak characterization in selected cases but are not mandatory for routine practice.
Keywords
Introduction
Endovascular aneurysm repair (EVAR) is a widely accepted, safe technique for treatment of aortic aneurysm disease. Patients after EVAR need lifetime follow-up due to stentgraft complications such as limb thrombosis, component disconnection, migration, and endoleaks provoking late aneurysm rupture [14, 30]. Nowadays, multi-slice computed tomography angiography (CTA) with early and late phase sequences is the most frequently used modality for follow-up after EVAR and for diagnosis of an endoleak [18, 30]. Guidelines of the Society for Vascular Surgery and the European Society for Vascular Surgery recommend CTA (and plain radiographs) at one and 12 months after EVAR, and an additional CTA at 6 months in case of type II-endoleak or other abnormalities [6, 18]. No endoleak and/or a stable or shrinking aneurysm diameter at 12 months after EVAR, leads to yearly Colour Duplex Ultrasound (CDU) (and plain radiographs) [6, 29]. Meanwhile, the repetitive use of CT scanning has raised concerns related to the use of nephrotoxic contrast agents, the cumulative radiation exposure and potential lifetime cancer risk [6]. The value of CEUS in detection and characterization of endoleaks after EVAR has already been demonstrated in several studies [11, 31]. So, CEUS is recommended with strong evidence for accurate endoleak detection and characterization according to the latest EFSUMB (European Federation of Societies for Ultrasound in Medicine and Biology) guidelines on non-hepatic applications 2011 [25]. However, ultrasound is considered very observer dependent in identifying endoleaks and underestimates aortic size compared to CTA [7, 27] with an inter-observer reproducibility range from±2 mm to±10 mm [16, 27].
Development of 3D/4D US and image fusion potentially offers improved measurement of aneurysm diameter compared to 2D techniques and a higher sensitivity in endoleak diagnostics [1–3, 15, 20].
Therefore, the primary endpoint of this study was to compare novel techniques like 3D/4D US and image fusion (combined real-time display of B mode and CT scan) with internally standardized 2D leading-edge US to measure the maximum aortic aneurysm diameter in routine follow-up after EVAR. Secondly, the advanced imaging technique of CEUS was evaluated for detection of potentialendoleaks.
Material and methods
Study design
Patients
The prospective study included a cohort of 20 consecutive EVAR follow-up patients examined within one week at the University Medical Center Regensburg. Seventeen were males and three females. Median age was 69 years (51–84 years.). Out of the 20 investigated patients, eleven patients had been treated electively for infrarenal aortic aneurysm or penetrating aortic ulcer and nine patients had been treated electively for thoraco-abdominal aortic aneurysm. Median aneurysm size before infrarenal EVAR was 5.0 cm (3.5–7.7 cm) and 5.5 cm (5.0–8.0 cm) before fenestrated EVAR (n = 7) or branched-EVAR (n = 2).
Median time after the EVAR procedure was 12 months (16 days - 48 months). The study complies with the principles outlined in the Declaration of Helsinki and was performed in accordance with the guidelines of Clinical Hemorheology and Microcirculation. The study was approved by the local research ethics committee (REC number 12-101-0121). Subjects gave informed consent for all procedures including CEUS.
Study protocol
Five different ultrasound techniques were used as follows. Measurements of the anteroposterior and lateral maximum aortic aneurysm diameters were done in triplets in a transverse view for each of the 2D, 3D, 3D Curefab™, 4D and image fusion modes. To avoid overestimation of the aneurysm size the transducer was moved from the largest diameter to a plane depicting a nearly circular-shaped structure. A leading-edge method was used, with measurement from the outer near wall to the inner echoes of the opposing aortic wall. Additionally CEUS was performed to detect and classify any endoleaks. The examination week was selected at random.
Routinely, US scans during follow-up are always combined with plain radiographs to detect stentgraft migration or stentgraft disconnection. Multi-slice CT scans, with an early and late phase, are performed: postoperatively, after 12 months, and in every patient with increasing aneurysm size, endoleak type I/III or with an new endoleak type II detected during follow-up, as published elsewhere [22]. However, both techniques are not object of the following study.
Ultrasound imaging
CDU and 2D CEUS
CDU and 2D CEUS were performed using the Hitachi HI VISION Ascendus (Hitachi Medical Systems Europe Holding AG, Zug, Switzerland), and EUP-C715 1–5 MHz convex multi-frequency sector transducer. For measurement of the maximum aneurysm sac diameter, 2D US was internally standardized as described earlier [26].
The standardized ultrasound examination was performed perpendicular to the aortic axis to create an almost circular cross-section of the aneurysm. The diameter was determined from the anterior to posterior and the left to right lateral walls using the “Leading-to-Leading edge” (LTL) method [7]. Sulphurhexafluorid (SonoVue®, Bracco Imaging Deutschland GmbH, Konstanz, Germany), 2 ml, was used for endoleak diagnosis by CEUS after informed oral and written consent (Fig. 1A). Important technical aspects of CEUS examinations have already been described elsewhere [10].
3D B mode and CEUS
Special software for the Hitachi HI VISION Ascendus is necessary to create a complete 3D B mode and contrast enhanced US volume data set with the EUP-C715 convex transducer by a free-hand sweep technique (Fig. 1B). In a transverse view, the probe is moved caudally with a sweep lasting between 8–10 seconds.
Magnetic field tracked 3D CEUS (Curefab™)
The CurefabTM system (Curefab Technologies GmbH, Munich, Germany) has already been described elsewhere [1]. In brief, it is an add-on system, which uses magnetic field tracked sensors installed on the ultrasound transducer (Hitachi HI VISION Ascendus - EUP-C715 1–5 MHz convex sector transducer).
The CurefabTM system is not feasible for patients with pacemakers because of the electromagnetic field generated. Imaging of the aortic aneurysm in 3D with a virtual centerline is comparable to the known post-processing CTA-programs like Osirix™ (Fig. 1C).
4D US
A volume data set was automatically generated in real-time using a volume transducer (Hitachi HI VISION Ascendus EUP-CV714 2–5 MHz probe) held in a fixed position on the abdominal wall and employing additional 4D software (Fig. 1D).
Image fusion
Image fusion was also performed with the Hitachi HI VISION Ascendus (EUP-C715 1–5 MHz probe) using a previously acquired, postoperative CTA (Fig. 1E). Additional hardware components required were an electromagnetic generator and tracking sensors clipped to the transducer [8]. Special software tracks the exact location of the transducer in real-time. Firstly, the CTA-DICOM data set was uploaded to the ultrasound machine. The postoperative CT (with or without contrast) was most suitable for fusion imaging because the endovascular stentgraft itself changed the aneurysm morphology in the follow-up. Quick and accurate co-registration of the CT- and ultrasound images (Fig. 1E) was achieved using the following characteristic registration points: In cross-sectional view, the navel and a typical stent structure (e.g. iliac leg) was fixed in a first plane, and the superior mesenteric artery (SMA) or celiac trunk were connected in a second plane. Image fusion is contraindicated in patients with pacemakers because of the electromagnetic field.
Statistical analysis
The standardized 2D method was defined as the gold standard. Measurement values for the five ultrasound methods were done in triplets and expressed as mean±SD in scatter and Bland-Altman plots. The Friedman test was used to compare them all. Intra-class correlation (ICC) was calculated. Statistical evaluations were performed using SPSS (IBM SPSS Statistics 22, Ehningen, Germany). Values of p < 0.05 were considered as statistically significant.
Results
Measurement of aortic ap- and lateral diameter by 2D US, 3D Curefab™, and 4D US and image fusion
The mean aortic ap-diameters were 4.0±1.3 cm (1.9–7.4 cm) for 2D US, 4.0±1.2 cm (2.0–7.4 cm) for 3D Curefab™, 3.9±1.3 cm (2.2–7.4 cm) for 4D US, and 4.0±1.2 (2.2–7.4 cm) for image fusion (Fig. 2A). Standardized 2D US was compared with 3D Curefab™, 4D and image fusion as shown in Bland-Altman (Fig. 2B) and scatter plots (Fig. 2C-E). Friedman test showed no significant differences between the four different methods (p > 0.05). Bland-Altman plots demonstrated a mean difference of 0.17 mm for 3D Curefab™, 0.93 mm for 4D US, and 0.02 mm for image fusion (agreement). The ICC was 0.99 for 2D US, 0.99 for 3D Curefab™, 0.98 for 4D US, and 0.99 for image fusion.
Mean lateral diameter measurements were 4.1±1.2 cm (2.1–7.0 cm) for 2D US, 4.0±1.1 cm (2.2–7.2 cm) for 3D Curefab™, 4.2±1.2 cm (2.3–7.6 cm) for 4D US, and 4.09±1.2 (1.8–7.6 cm) for image fusion (Fig. 3A). Again, the mean aortic lateral diameters of every patient measured by 3D Curefab™, 4D US and image fusion were compared with 2D measurements. Friedman test showed no significant differences between the four different methods (p > 0.05) but only ICC of 0.94 for 2D US, 0.94 for 3D Curefab™, 0.94 for 4D, and 0.90 for image fusion. Bland-Altman (Fig. 3B) and scatter plots (Fig. 3C-E) demonstrated a mean difference of 0.87 mm for 3D Curefab™ and 0.03 for image fusion (agreement), and 1.43 for 4D US.
Estimation of aneurysm growth in follow-up
Out of the 20 patients, 17 (85%) had clearly shrinking aneurysm size by 2D, 3D, 4D and image fusion ultrasound measurements. Of the remaining three patients, one had suspected increasing aneurysm size by all four methods and one to have constant aneurysm size. Therefore consensus between the four different ultrasound methods was found in 19 of the 20 patients (95%). The third patient was estimated to have constant aneurysm size by 2D US and 3D Curefab™ but shrinking aneurysm size by 4D US. The definitive decision could be made by image fusion comparing actual ultrasound image and postoperative CT scan in real-time (Fig. 4).
Measurement of aortic volume by 3D Curefab™ and 4D US
The mean aortic volume evaluated by free-hand but magnetic field tracked 3D Curefab™ and 4D US was 84.4±59.8 ml (17.6–276.7 ml) and 59.4±48.0 ml (8.4–190.0 ml), respectively. The mean difference between the two methods was 24.0 ml±31.0 ml with significantly larger values measured by the free-hand 3D Curefab™ (p < 0.05). However, the volume data set was generated automatically by the 4D technique. Consequently different aortic lengths had been used for volume calculation in the two methods. Aortic lengths were 7.1±2.5 cm for 3D Curefab™ and 5.8±1.9 cm for 4D (p < 0.05). Standardizing both methods to a comparable aortic length, a linear correlation was observed between 3D Curefab™ and 4D (correlation coefficient = 0.86, p < 0.05) (Fig. 5).
Endoleak detection
Only two out of the 20 patients (10%) in this one-week study cohort had an endoleak type II. Whereas the persistent lumbar endoleak type II of the first patient had also been identified by 2D CEUS, endoleak characterization of the second patient was more complex. After receiving a fenestrated stentgraft, growing aneurysm size and an endoleak type II in follow-up prompted embolization with an ethylenvinyl alcohol polymer in dimethylsulfoxid (Onyx® LES) for the lumbars and coiling of the inferior mesenteric artery (IMA). Post-interventional CTA was not able to detect an endoleak because of the huge extinction artefact (Fig. 6A). However, increasing aortic diameter indicated re-intervention. Whereas 2D CEUS showed a strong endoleak type II, probably from lumbar arteries, further feeding arteries remained unclear (Fig. 6B). In this special case, free-hand, magnetic field tracked 3D as well as image fusion localized the supposedly occluded IMA as an endoleak-feeding artery (Fig. 6C-E).
Discussion
CTA is the most accepted and widely performed imaging after EVAR in vascular surgery although the cumulative radiation dose and potential nephrotoxicity limit their use in older aged patients in lifelong surveillance. In contrast, duplex ultrasound, especially CEUS, is a developing technique strongly recommended and evaluated by comparison studies with CTA [4, 24]. At present, restricted technical and personnel resources hinder rapid and wide application despite the evident advantages. The key objectives of follow-up after EVAR are firstly, the maximum aneurysm size and secondly, endoleak detection [22]. CTA is considered the gold standard while having an inter-operator variability of more than 5 mm in 17% of the cases [26]. Ultrasound has not been generally standardized up to now, accepting different methods of measurement: inner-to-inner (ITI), leading-to-leading edge (LTL), and outer-to-outer (OTO) diameters [7]. Consequently, several studies showed discrepancies in the different ultrasound methods between 0.17 and 0.39 cm and underestimated the aortic diameter compared to CTA [7, 27]. Thus, correct positioning of the probe to visualise a circular-shaped structure plays an important role in achieving comparable results of a kinked and elongated aorta [26]. With new 3D/4D ultrasound techniques and image fusion available [23], aortic aneurysm diameter measurement has been again investigated and the benefit for endoleak detection re-evaluated.
Therefore, primarily, our study protocol that matched different ultrasound methods for measuring aneurysm diameter was internally standardized with a convex probe in a rotated position and using a 2D cross-sectional leading-edge measurement. With this technique mean differences for 2D US and 3D/4D US or image fusion ranged between the low values of 0.2 and 1.4 mm. The ICC of standardized 2D and the advanced US imaging was excellent with 0.99 for the anteroposterior diameter, and good, with 0.94 for the lateral diameter. Delineation in the axial direction in ultrasound is well known to be superior to depiction of lateral boundaries because of the higher acoustic impedance mismatch [26]. The Friedman Test showed full agreement between 2D US and magnetic field tracked 3D/4D US, and image fusion ap-measurements. In conclusion, aortic aneurysm size was correctly identified in 19 out of 20 patients (95%). In only one case, image fusion was required to make the definitive decision. Currently, 2D measurements are sufficient for routine practice in follow-up of maximum aortic aneurysm diameter but advanced US techniques are able to provide new insights. Volume measurements are believed to be the better prognostic parameter in aneurysmal growth, independent of whether automatically performed by a special probe or in a magnetic field tracked free-hand way [2]. Good short-term results were achieved in infrarenal aneurysms with predetermined boundaries between the renals and the aortic bifurcation, and perfect visualization [9, 15]. Otherwise, in pararenal or thoracoabdominal plus iliac aneuryms, it might be difficult to standardize the aortic length. In our experience with 20 patients, volume calculations were very heterogeneous. Surprisingly, fixed aortic length correlated well between 3D and 4D US, comparable to Long A et al and their partial volume estimation [15].
CDU is considered to be less accurate than CTA in characterisation of endoleaks [17]. A large meta-analysis (21 studies in 2601 patients) of Mirza TA et al showed a sensitivity of 0.77 and a specificity of 0.94 versus 0.98 and 0.88 [17]. The same meta-analysis compared CEUS and CTA (7 studies, 288 patients) and found high sensitivity of 0.98 and an adequate specificity of 0.88 [17]. Recently, these results were approved by Gurtler VM et al [12] for infrarenal EVAR and Perini P et al. for fenestrated stentgrafts [21]. Subsequently, both of them suggested switching from CTA to CEUS in long-term surveillance [12, 21]. The first follow-up protocols have already been published which argue that an initial postoperative CTA is not necessary for most patients [19]. Meanwhile 2D CEUS has the highest level of evidence and is strongly recommended for the detection and characterization of endoleaks in the ultrasound society [25]. 3D CEUS might be even more sensitive to detect endoleaks following EVAR than 2D CEUS, DSA or CTA [1, 20]. Image fusion showed good visibility of even small endoleaks [8].
In our study, 2D CEUS was absolutely certain to detect endoleaks but unreliable in their classification. As expected, 3D/4D CEUS and image fusion obviously improved visualization and characterization of endoleak feeding arteries in selected cases especially after intervention that creates artefacts. But according to our experience, these advanced techniques are not mandatory in every day practice.
Our investigation is limited because of several preconditions. In fact, CEUS is a semi-invasive method with a very low complication rate owing to pseudoallergic reactions but needs written informed consent, an experienced examiner and an intravenous line as well as personal assistance. Digital loops has to be recorded to re-evaluate the results. Therefore costs for imaging storage, special 3D, 4D and image fusion software just as technical equipment for magnetic field tracking has to be taken into account. A high resolution CT scanner is necessary to receive suitable scans for connecting with ultrasound in image fusion technique.
On the other hand strength of the study is the back-to-back application of 5 different ultrasound techniques (2D, 3D free-hand and magnetic field tracked Curefab™, 4D, image fusion) with three repetitive measurements by one experienced examiner resulting in a total of 300 values.
Conclusion
Novel ultrasonographic imaging techniques as 3D/4D US and image fusion potentially offer improved measurement of aneurysm diameter compared to 2D techniques as well as a higher sensitivity in endoleak diagnostics. In our one-week study, 20 patients underwent five different US techniques (2D, 3D free-hand and magnetic field tracked Curefab™, 4D, image fusion) with three repetitive measurements made by one experienced examiner resulting in 300 values. Statistical analysis demonstrated almost similar findings. Standardized 2D US allowed satisfactory routine follow-up of maximum aortic aneurysm diameter after EVAR and compared to advanced imaging techniques such as 3D/4D US and image fusion. Image fusion exclusively made a definitive statement about aneurysm growth without the need for new CTA imaging using a direct comparison of the postoperative CT scan with B mode imaging in a real-time dual-image display. 3D/4D CEUS and image fusion enabled improved endoleak characterization, but only in selected cases. According to our experience, these new imaging methods are not mandatory for daily practice.
Funding
This study was in part supported by the Bavarian government (”Leitprojekte Medizintechnik“, Kapitel 0703, Titel 68668).
Conflict of interest
PM Kasprzak, H Apfelbeck, C Hennersperger, and EM Jung declare no conflicts of interest.
W Schierling received travel costs from Bracco Imaging Deutschland GmbH.
K Pfister received travel costs and speaker’s fee from Bracco Imaging Deutschland GmbH and speaker’s fee from Hitachi Medical Systems.
Presentation information
This study was in parts presented at the 18. Chirurgische Forschungstage, Hannover, Germany, 09.-11.10.2014 and at the 30. Jahrestagung der Deutschen Gesellschaft fuer Gefaesschirurgie und Gefaessmedizin, Hamburg, 24.-27.09.2014. Both presentations were in German.
Footnotes
Acknowledgments
The authors like to thank Christian Buchner (Hitachi Medical Systems) and Curefab Technologies GmbH for technical assistance as well as Katrin Thelen for statistical assistance.