Wall shear stress and relative residence time as potential risk factors for abdominal aortic aneurysms in males: a 4D flow cardiovascular magnetic resonance case–control study

18 AAA male patients (66–76 years) and 22 age- and sex-matched controls were recruited from an AAA screening program at Linköping University Hospital. AAA patients presented with maximum abdominal aortic diameters ≥ 3.5 cm, while matched controls had no focal dilation and diameters < 2.5 cm in the abdominal aorta on ultrasound. Additionally, 23 sex-matched young (18–30 years) subjects were enrolled. Inclusion criteria included sinus rhythm, and no contraindications for CMR. Characteristics for the three study cohorts are summarized in Table 1. The study was approved by the regional ethical review board and all subjects gave written informed consent.

All subjects underwent a CMR exam at a 3T CMR system (Ingenia, Philips Healthcare, Best, The Netherlands) equipped with a 32-channel torso coil with 60 cm coverage. Balanced steady-state free precession (bSSFP) images were acquired and used to assess aneurysm and thrombus morphology. Scan parameters included: axial slab covering the abdominal aorta, flip angle 60°, echo time 1.5 ms, repetition time 3.1 ms, and spatial resolution 1.4 × 1.4 × 5 mm. AAA morphology was visually determined by one observer (Table 2). Contrast-enhanced CMR angiography (CE-CMRA) volumes were acquired following administration of a gadolinium contrast agent (Magnevist, Bayer Healthcare, Berlin, Germany) and used to determine relevant aortic landmarks. Scan parameters included: a sagittal-oblique slab set to cover the whole aorta from left ventricular outflow tract to iliac bifurcation, flip angle 30°, echo time 2 ms, repetition time 5.8 ms, parallel imaging (SENSE) factor 4, and spatial resolution 0.6 × 0.6 × 1.2 mm. 4D flow CMR data were acquired with a free-breathing, respiratory navigator gated, retrospectively cardiac-gated sequence immediately after the CE-CMRA. The 4D flow sequence used a center-out acquisition strategy to acquire the central parts of k-space while there was still contrast agent in the blood. Scan parameters included: a sagittal-oblique slab with 3D field of view (FOV) = 480 – 560 × 480 – 560 × 71 – 117 mm³ and matrix size = 192 – 224 × 192 – 224 × 28 – 46 adjusted depending on each subject’s anatomy to cover the whole aorta from the aortic valve to the iliac bifurcation, velocity encoding range (VENC) 100–200 cm/s, flip angle 15°, echo time = 2.5–3.1 ms, repetition time = 4.4–5.0 ms, k-space segmentation factor 2, effectively acquired temporal resolution 35–40 ms, and acquired spatial resolution 2.5 × 2.5 × 2.5 mm³. SENSE parallel imaging was used, with a total acceleration factor of 4–4.8. Data were reconstructed to 40 timeframes. Total scan time including respiratory navigator gating was 10–15 min. 4D flow datasets were corrected for concomitant gradient field effects on the scanner, while phase-wraps and background phase-offset errors were corrected offline [25, 26]. Background phase-offset errors were corrected using a weighted 4th order fit to static tissue [26].

Phase-contrast CMR (PC-CMR) images were computed from the magnitude and velocity images at peak systole, defined as the timeframe with maximum mean velocity in the whole aorta. 3D segmentation was performed using ITK-snap with a semi-automatic region growing algorithm combined with manual adjustments [27]. The 3D segmentation was registered to every other timeframe in the 4D flow CMR data by using a non-rigid registration method based on the Morphon algorithm. The Morphon algorithm uses diffeomorphic field accumulation together with fluid and elastic regularization of the displacement fields in order to generate physically plausible deformations [28]. Consequently, isosurfaces at each timeframe were generated and registered to the peak systolic isosurface with a non-rigid variant of the iterative closest point algorithm [29]. In this way, the same node on the peak systolic isosurface could be tracked on isosurfaces at each timeframe, allowing for computation of time dependent parameters.

Instantaneous WSS vectors were computed with a previously described method [30].The WSS computation was conducted using three points on the inward normal vector and a length of the vector itself of 12 mm, 10 mm and 9 mm for AAA patients, elderly controls and young controls, respectively. This cohort-specific inward vector took into account the differences in aortic lumen size between the cohorts, and fall within the optimum range as defined by Potters et al. [30]. A Carreu-Yasuda model was used for viscosity with viscosity at infinite shear rate assumed to be 0.0035 [Pa s]. Based on the instantaneous WSS vectors, the time-averaged wall shear stress (TAWSS) [Pa] through cardiac cycle were computed as:

Further, the oscillatory shear index (OSI) [−] was computed as

Finally, the relative residence time (RRT)[−], which estimates the relative duration that blood resides close to the wall through a combination of TAWSS and OSI, was computed as in [14]

The aorta was divided into five segments and WSS parameters were computed in each segment. First, five landmarks were determined by visual inspection of one expert observer on CE-CMRA images: aortic valve, brachiocephalic trunk and left subclavian insertion on the arch, renal arteries and iliac bifurcation. These annotations were registered to the 4D flow images, projected on the centreline of the aorta, and subsequently used to define five segments (Fig. 1): ascending aorta (AAo)—from the valve to brachiocephalic trunk; arch—from brachiocephalic trunk to the point 20 mm distal from the left subclavian artery; descending aorta (DAo)—from the distal end of the arch to midway between the end of arch and the renal arteries; suprarenal abdominal aorta—from the distal end of DAo to renal arteries; infrarenal abdominal aorta—from the renal arteries to the iliac bifurcation. The time frame with peak flow rate (TF_peak) were identified for each segment [31]. Average velocity and average WSS at TF_peak (peak velocity and peak WSS, respectively) were computed in each segment. Additionally, mean TAWSS, OSI, and RRT were computed for each segment. Finally, the maximum diameter normalized by body surface area (BSA), was obtained for each segment based on the cross-sectional area.

Further, a local analysis was performed to identify the extent to which the infrarenal abdominal aorta was exposed to abnormal WSS. Abnormally low peak WSS was defined as WSS values more than two standard deviations lower than the mean WSS in the elderly cohort. Similarly, abnormally high OSI and RRT were defined as values more than two standard deviations higher than the mean values in the elderly cohort. The area affected by abnormal values was computed and compared to the elderly cohort.

Finally, to assess the impact of the intraluminal thrombus on WSS parameters, the intraluminal thrombus was segmented on the bSSFP images by one operator on all 13 AAA patients in which the intraluminal thrombus was visually identifiable. The bSSFP images were registered to the 4D Flow images to identify the interface between the lumen and intraluminal thrombus. Average values of peak WSS, TAWSS, OSI and RRT acting on the entire infrarenal abdominal aorta and the thrombus-free wall in the infrarenal abdominal aorta were compared.

The Shapiro–Wilk test was used to test the normality of the distribution of each parameter in each segment. Results were expressed by means of mean ± (standard deviation) when normally distributed and median [interquartile range] otherwise. Moreover, when normally distributed, a two-tailed unpaired Student’s t-test was used to compare AAA vs elderly controls, and young controls vs elderly controls. Bonferroni correction for multiple comparison was applied and a p value < 0.025 (0.05/2) was considered significant. If not normally distributed, a non-parametric Wilcoxon rank sum test was used. Two-tailed unpaired student’s t-test was also used to compare values between the infrarenal abdominal aorta and suprarenal abdominal aorta, when data were both normally distributed, and a non-parametric Wilcoxon rank sum test was used when data were not normally distributed.

AAA patients and elderly presented with similar diameters, peak velocities, and peak WSS in the AAo and the arch, whereas AAA had higher diameters in the abdominal aorta (p ≤ 0.001 in suprarenal abdominal aorta and < 0.001 in IAA), and lower velocities (p = 0.0053 in DAo, p ≤ 0.001 in suprarenal abdominal aorta and < 0.001 in infrarenal abdominal aorta), as well as lower peak WSS (p < 0.001 in DAo, suprarenal abdominal aorta and infrarenal abdominal aorta) in the descending and abdominal aorta. TAWSS were lower in AAA in each segment (p ≤ 0.001 in AAo, p = 0.0031 in arch and < 0.001 in DAo, suprarenal abdominal aorta and infrarenal abdominal aorta). AAA patients and elderly controls had similar OSI in the whole aorta, except in the DAo (p = 0.0047). RRT is similar between the two groups in the arch, whereas AAA patients have higher RRT (p = 0.0023 in AAo and p < 0.001 in DAo, suprarenal abdominal aorta and infrarenal abdominal aorta) in the rest of the aorta (Table 3 first and second column, and Fig. 2 red and blue graphs).

While in elderly controls peak WSS and TAWSS values are constant in whole descending aorta, peak WSS and TAWSS decrease from suprarenal abdominal aorta to infrarenal abdominal aorta in AAA patients (p < 0.001). OSI and RRT increase from suprarenal abdominal aorta to infrarenal abdominal aorta in AAA patients (p < 0.001), as well as in elderly controls (p < 0.001).

Elderly controls had larger diameters, lower peak velocities, lower peak WSS, and lower TAWSS values than young subjects in each segment (p ≤ 0.001). Moreover, elderly controls had higher OSI in all segments except in the infrarenal abdominal aorta (p ≤ 0.001). Elderly controls had almost doubled RRT compared to young controls in all segments, except for the infrarenal abdominal aorta (p < 0.001) (Table 3 second and third column, and Fig. 2 blue and green graphs).

In elderly controls peak WSS and TAWSS mean values were similar in the suprarenal abdominal aorta and infrarenal abdominal aorta, while in young controls, peak WSS and TAWSS were higher in the suprarenal abdominal aorta when compared to the infrarenal abdominal aorta (p < 0.001). In both elderly and young controls, OSI and RRT were higher in the infrarenal abdominal aorta when compared to the suprarenal abdominal aorta (p < 0.001).

Blood flow in infrarenal abdominal aorta was visualized with streamlines for one AAA (left), one elderly control (central) and one young control (right) in Additional file 1. In the AAA a vortex is visible during diastole, which is not present in the abdominal aorta of the controls. For the same subjects, 3D WSS maps in the infrarenal abdominal aorta over the cardiac cycle are shown in Additional file 2.

The mean peak WSS value minus 2 standard deviations in the elderly controls was 0.57 Pa, while mean OSI plus 2 standard deviations and mean RRT plus 2 standard deviations were 0.36 and 18.37, respectively. The percentage of infrarenal abdominal aorta vessel wall area exposed by abnormally low peak WSS was 0.25 [1.2] % in elderly controls and 24.7 [5.9] % in AAA (p < 0.001). The exposure of abnormally high OSI in the infrarenal abdominal aorta was lower in elderly controls than AAA patients (2.3 [3.6] % vs 1.2 [2.1] %, p = 0.0074). The infrarenal abdominal aorta in the AAA patients was more exposed to abnormally high RRT than in the elderly controls (10.6 [7.9] % vs 1.4 [2.7] %, p < 0.001), but there was no difference between elderly and young controls. Sites with elevated RRT were located mainly in posterior infrarenal abdominal aorta in young and elderly controls, whereas regions with high RRT are located inside the aneurysm for AAA patients (Fig. 3).

Figure 4 shows peak WSS, TAWSS, OSI, and RRT acting on the whole infrarenal abdominal aorta and on the thrombus free vessel wall, respectively, for the 13 AAA patients with an identifiable intraluminal thrombus. For these 13 AAA patients, including or excluding the surface area with intraluminal thrombus did not significantly impact the computation of peak WSS, TAWSS, OSI, or RRT.

This study has several limitations. First, the number of AAA patients included is relatively small, so the statistical power of the study is limited. Moreover, inclusion of more patients would have enabled comparisons between different aneurysm shapes, such as fusiform vs saccular. Nevertheless, this number of AAA patients included in this study is larger than in previous studies of hemodynamics in the abdominal aorta. Second, follow-up data on AAA growth and rupture were not available, thus limiting the study to a cross-sectional design. Therefore, it is not possible to determine if alterations in RRT are simply correlated to the presence of AAA or if they could contribute to the development of AAA. Future work includes a longitudinal follow-up which may lead to new insights into the role of the hemodynamics markers studied here. Finally, WSS estimation with 4D flow CMR is limited by spatial resolution effects (30, 38). In particular, the estimated WSS depends on the spatial resolution. Another spatial resolution limitation of the WSS algorithm used here is that at least 8 voxels across the diameter of the targeted vessel are recommended for accurate WSS estimates. This does not hold for the infrarenal abdominal aorta in the young subjects in this study.