3D black blood VISTA vessel wall cardiovascular magnetic resonance of the thoracic aorta wall in young, healthy adults: reproducibility and implications for efficacy trial sample sizes: a cross-sectional study

Cardiovascular disease (CVD) remains a major cause of morbidity and mortality worldwide [1]. Atherosclerosis, the underlying pathophysiological condition, is a slowly progressive, inflammatory process that causes adverse remodelling of the arterial wall. The slowly progressive changes in the arterial wall attributed to atherosclerosis provide a window of opportunity for its pre-clinical detection.

Cardiovascular magnetic resonance (CMR) has evolved as an attractive imaging modality for studying atherosclerosis in vivo [2‐4]. The descending and abdominal aorta, the artery where atherosclerosis is known to develop first [5], has been the subject of a number of studies using MR [6‐8]. To date, most CMR studies have used 2D sequences at a field strength of 1.5 tesla (T), limiting both spatial resolution and aortic coverage due to time constraints. Nowadays, the widespread availability of 3 T MR scanners has led to increased signal-to-noise ratio vessel wall imaging [9]. Furthermore, there is ongoing development of sequences that aim to optimize image quality whilst covering a larger extent of the aorta with clinically acceptable scan duration [10‐12].

One of the most promising CMR sequences that is increasingly being applied in various vascular areas is a 3-dimensional (3D), T1-weighted (T1), black-blood (BB), turbo-spin-echo (TSE) sequence with variable flip angles (3D-T1-BB-VISTA) [6, 13]. This sequence allows for an isotropic depiction of the vessel wall in an acceptable scan time with (sub)-millimetre spatial resolution. Because of the T1-weighting, the VISTA sequence is also sensitive to the effects of contrast agent administration [13]. Hence, the 3D-T1-BB-VISTA sequence may be attractive for the detection of atherosclerosis in early life, when vascular wall abnormalities are still subtle.

The reproducibility of the 3D-T1-BB-VISTA sequence has been evaluated before, yet in an older population with known CVD and using CMR hardware from a different vendor as compared to the current study [6]. The objective of this study was to investigate the inter-scan and inter-rater as well as intra-rater reproducibility of 3D-T1-BB-VISTA CMR in healthy, young adults for assessment of vascular wall characteristics of the descending thoracic aorta. Based on the reproducibility, sample sizes for (therapeutic) trials were calculated by determining the number of participants needed to detect pre-specified differences in aortic wall characteristics between groups.

The present study demonstrated that the 3D-T1-BB-VISTA sequence is highly reproducible and well-suited for quantification of aortic wall characteristics in young adults. The inter-scan, inter-rater and intra-rater limits of agreement of all aortic wall characteristics under study were narrow and the ICCs were all above 0.75, indicating a low variability in measurement. Sample sizes required to detect a 5 % difference between 2 groups were reasonable for all of the studied aortic wall characteristics. Our results provide an important first step in the process of determining the utility of the 3D-T1-BB-VISTA sequence for preclinical detection and monitoring of atherosclerosis burden in early adult life, when vascular wall abnormalities are still subtle.

The gradually developing lesions in the arterial wall, induced by atherosclerosis, offer an opening for its pre-clinical detection. Unfortunately, currently used CV risk scores are population based, yet do not conclusively identify individuals at high-risk for developing clinically manifest CVD [20]. Furthermore, they are predominantly validated in people over 40 years of age [21]. The American Heart Association (AHA) has recognized this problem and underlined the need for enhanced assessment of non-invasive methods for CVD risk evaluation in young adults [21]. More detailed assessment of the presence or absence of early atherosclerotic changes may improve our understanding of the pathophysiology of atherosclerosis, reveal important determinants and may enhance CVD risk stratification, which is important since identification of high-risk individuals at a young age may enable early interventions to avert or delay the onset of clinically manifest CVD.

Aortic atherosclerosis develops at an early age with the occurrence of aortic wall thickening, a process that precedes the development of discrete plaques by up to several decades. The extent of aortic wall thickening and plaque burden are known to be related to CV risk factors and events [2, 4, 8, 17‐19, 22]. Moreover, characteristics of the descending thoracic aortic wall highly relate with those of the carotid arteries [8, 23]. Hence, aortic atherosclerosis is conceived to be a relevant proxy measure of generalized atherosclerosis and possibly of specific interest in young adults.

The assessment of aortic atherosclerosis using imaging modalities is promising [2]. Of all the available modalities for this purpose CMR has various advantages over other imaging techniques. CMR is non-invasive, does not rely on ionizing radiation, and is capable of large anatomical coverage and superior soft tissue contrast that offers the ability to directly visualize and accurately assess various stages of aortic atherosclerosis. Prior studies have demonstrated a good agreement between CMR and histology as well as transesophageal echocardiography with regard to the presence and extent of atherosclerotic plaques [2, 3, 8, 23]. Furthermore, Corti and co-authors suggested, based on their reproducibility data, that 2D CMR is capable of precisely measuring subtle alterations in aortic lesions over time; a change of >5 % in plaque size can be accurately detected [2, 19].

Given the abundance of different CMR sequences, it is desirable to reach consensus on which CMR sequence is most appropriate for early identification and subsequent monitoring of aortic atherosclerosis burden [6‐8, 10‐12, 22]. Currently, there is an ongoing development of CMR sequences that aim to optimize aortic atherosclerosis quantification. A number of these sequences have shown to be feasible for the quantification of aortic atherosclerosis burden in a highly reproducible manner in healthy subjects and in subjects with CVD [6‐8, 10‐12, 22]. To date 2D Double Inversion Recovery (DIR) imaging remains the most commonly used CMR technique for evaluation of the aortic wall [11, 24]. The majority of aortic CMR studies and proposed sequences to date have been performed at 1.5 T using 2D BB techniques and a slice thickness of 4-5 mm. Additionally, they provided limited aortic coverage due to time constraints [7, 8, 10‐12, 22, 24]. The present study was performed at a 3.0 T system prepared with a dual-source RF-transmission system, allowing for a higher SNR than a 1.5 T system and thus increases spatial resolution [25]. Additionally, this study used a 3D BB sequence. Prior studies have shown that 2D and 3D sequences provide similar morphometric measurements of the aorta. These studies suggest that major benefits of 3D imaging as compared to 2D imaging are the rise in SNR, the ability to reformat the data in any desired orientation with MPR, more detailed coverage of the arterial wall, the reduction in examination duration, the augmented visualization of small plaque constituents and the improved reproducibility (due to smaller slice thicknesses and enlarged flexibility in co-image registration when performing a sequential study) [6, 12, 26‐28]. Moreover, a slice thickness of 1.2 mm was used, thereby reducing partial volume effects and providing a high isotropic resolution, which may reduce variability in measurement as compared to sequences that use a larger slice thickness [8]. Hence, in contrast to prior studies, the 3D-T1-BB-VISTA sequence is likely to improve visualization and delineation of the aortic wall and as such allows for an enhanced accuracy and quantification of measurement. In addition, the studied sequence is capable of covering approximately 35 cm of aorta within an acceptable scan time, enabling us to assess its reproducibility, and thus possibly aortic atherosclerosis, over a larger aortic territory and a wide range of aortic sizes. Another advantage of the 3D-T1-BB-VISTA sequence is that it can be used before and after contrast agent administration and thus can directly compare signal intensity before and after contrast agent administration. As such, it may be of value for evaluation of plaque burden, composition and neovascularization [2, 29]. In this study we demonstrated no relevant differences in arterial wall area and thickness between the non-contrast enhanced and contrast-enhanced acquisitions, implying that slow flow artefacts, which possibly decrease the accuracy of measurement, may not be an issue when using the 3D-T1-BB-VISTA sequence in healthy subjects. Yet, this observation does require further evaluation in other cohorts. In addition, a direct comparison of the reproducibility between the 3D-T1-BB-VISTA sequence and 2D sequences in young, healthy individuals remains to be performed.

Mihai and co-workers demonstrated an excellent reproducibility of the 3D-T1-BB-VISTA sequence in 15 patients with CVD and 6 healthy volunteers [6]. Our study extends these preliminary results to a much younger population. As opposed to Mihai and co-workers, the present study did not evaluate the reproducibility of the 3D-T1-BB-VISTA sequence at other locations than the thoracic descending aorta. The reasons for this were that atherosclerosis is known to develop first in the descending aorta and that the SNR of the region below the celiac trunk was insufficient when using a cardiac coil. Yet, since atherosclerosis burden may exhibit different characteristics in other vascular beds, the reproducibility of the 3D-T1-BB-VISTA sequence in the young at other locations than the thoracic descending aorta remains to be determined. Another difference is that the present study was solely performed in healthy young adults, who aortas are not likely to already harbour large atherosclerotic plaques. In contrast, our results confirm the ability of 3D-T1-BB-VISTA to detect subtle differences in aortic wall thickness between subjects with excellent reproducibility. Our findings are in accordance with the results of Mihai and co-workers, who demonstrated an excellent reproducibility of the 3D-T1-BB-VISTA sequence in patients with known CVD [6].

Our study has limitations. First, we did not use ECG triggering or breath holding. Hence, pulsation and breathing artefacts may have introduced slight errors in measurement of absolute aortic wall characteristics. Despite this limitation, reproducibility remained excellent. Nevertheless, tracking of motion using navigators or special motion compensation methods may also improve 3D sequences further. Second, due to time constraints, we analysed one out of every eight slices acquired, thereby inducing a wide gap between two analysed slices. As such, possible focal atherosclerotic lesions may have remained undetected. Yet, since atherosclerosis is conceived to be a generalized disease, we expect the potential loss of information to be minimal in young asymptomatic subjects. Third, our choice of slice thickness created some challenges in preserving adequate signal to allow continuous distinction of aortic wall boundaries in all slices. However, we specifically chose not to use thicker slices since prior studies on 3D BB aorta imaging also used slice thicknesses of ~1 mm with good results [6, 12, 26]. In addition, no slices had to be removed because of insufficient image quality and reproducibility remained excellent between scans as well as between and within raters. Moreover, we believe that the large size of our dataset, our 3D sequence and 3 T system have reduced the potential negative influence of our small slice thickness. Fourth, we did not use a fully automatic image analysis tool for the measurement of the aortic wall characteristics, which made quantification time-consuming and possibly less accurate. Yet, since vascular wall characteristics are usually in the range of a few millimetres, accurate quantification is crucial for its utility in atherosclerosis evaluation. The currently used semi-automatic tool requires dedicated training of raters in image analysis to achieve accurate quantification of aortic wall characteristics, particularly when vascular wall delineation between the vascular wall adventitia and enclosing tissue is unclear. Employment of standardized methods and fully automatic tools may increase efficiency and accuracy of quantification, thereby enabling atherosclerosis detection and monitoring over a larger territory and more homogeneity in measurement across studies. Finally, our results provide an important first step in the process of determining the utility of the 3D-T1-BB-VISTA sequence for preclinical detection and monitoring of atherosclerosis burden in early adult life, when vascular wall abnormalities are still subtle. However, the relation of the studied characteristics with CV risk factors and CVD morbidity and mortality remains to be determined.

In conclusion 3D-T1-BB-VISTA sequence provides excellent reproducibility for quantification of aortic wall characteristics and can detect differences of 5 % in aortic wall characteristics between groups with a reasonable sample size. It may be a valuable tool for assessment of the subtle vascular wall changes of early atherosclerosis.