This prospective single-center study was approved by our institutional review board. Written informed consent was obtained from each participant. Between June 2014 and December 2018, we enrolled 52 consecutive patients, including 18 first diagnosed and 34 revisited patients with TAK in this study. The definitive diagnosis of TAK was based on the 1990 American College of Rheumatology criteria [14]. TAK activity was examined according to the ITAS2010 [4]. All patients had been treated with glucocorticoids and immunosuppressive agents. High dose glucocorticoid therapy (40–60 mg/day) was used for induction of remission in the active TAK patients. Once remission occurred, the glucocorticoid dose was tailored to a target dose of 15–20 mg/day within 2–3 months and to $\le$10 mg/day after 1 year. According to clinical data and routine imaging findings, including those of ultrasonography, CTA, and MRI, the patients with TAK were classified into Types I to V [15]. We excluded nine patients because of contraindications to MRI (n = 3), contraindications to contrast media (n = 2), and poor image quality due to arrhythmia (n = 3) and poor holding of breath (n = 1). Fig. 1 depicts the study flowchart.

Fig. 1.

Flowchart for the selection and analysis of patients with TAK (n = 52). After excluding 9 patients, a total of 52 cases with TAK were included in the analysis.

All MRI examinations were performed using a 3.0 T MRI scanner (Verio, Siemens, Erlangen, Germany) equipped with a 32-channel surface-phased-array coil. The coil was placed beneath the mandible bone encompassing the whole thorax. As shown in Fig. 2, a 30-min comprehensive aortic MRI protocol was designed for the patients with TAK including dark-blood T2WI; contrast-enhanced magnetic resonance angiography (CE-MRA); pre, early, and delayed gadolinium enhanced T1-weighted imaging (T1WI); and 4D flow sequence.

Fig. 2.

Comprehensive aortic MRI protocol for patients with TAK. Pre-enhanced T1WI and dark-blood T2WI with transversal orientation performed after localization. CE-MRA was performed using the bolus-tracking sequence after contrast injection. Early and late enhanced T1WI were performed 1 and 15 min after contrast injection, respectively, with the repetition of pre-enhanced T1WI. Within the delayed period, a 4D flow sequence was performed for a duration of 6–9 min.

Dark-blood T2WI was performed in the axis orientation with a fat sat turbo spin echo (TSE) sequence (thickness = 5 mm, slice gap = 0 mm, time to echo/time to repetition (TE/TR) = 79.0 ms/4891.9 ms, field of view (FOV) = 38 $\times$ 28.5 cm${}^2$, and matrix = 320 $\times$ 168). Electrocardiography (ECG) and breathing trigger were adopted to minimize cardiac and respiration artifacts.

A breath-holding 3D gradient-echo (fast low-angle shot, FLASH) sequence (TE/TR = 1.1 ms/3.1 ms, FOV = 38.0 $\times$ 30.8 cm${}^2$, matrix = 384 $\times$ 281, slice thickness = 1.2 mm, and 104 slices per slab) was used to acquire CE-MRA in coronal orientation before and after administering the contrast medium. K-space was filled by adopting the central first approach. A postcontrast scan was performed after the intravenous administration of gadopentetate dimeglumine (Magnevist, Bayer, 0.2 mmol/kg body weight, 2 mL/s) with 30 mL of saline flush at the same injection rate. The optimal acquisition time was determined using a bolus-tracking sequence.

For T1WI, a three-dimensional fat-sat-gradient recall echo volumetric interpolated breath-hold examination (VIBE; TE/TR = 1.4 ms/3.9 ms, FOV = 38.0 $\times$ 30.8 cm${}^2$, matrix = 320 $\times$ 182, slice thickness = 3.5 mm, and 72 slices per slab) was performed in axis orientation. Early and delay enhanced images were acquired 1 and 15 min after contrast injection, respectively.

The 4D flow sequence covered the whole thoracic aorta with oblique coronal orientation in prospectively ECG-gated and navigation techniques (temporal resolution = 47.2 ms, slice thickness = 1.8 mm, TE/TR = 2.8 ms/47.2 ms, FOV = 34.0 $\times$ 25.5 cm${}^2$, and matrix = 192 $\times$ 115). Subsequently, a two-dimensional phase-contrast cine MRI was performed to acquire optimal velocity encoding (VENC) before the 4D flow sequence. The VENC value was 100–200 cm/s, which was determined using the maximum flow velocity measured in the thoracic aortic lumen. The acquisition time for the 4D flow sequence was 6–9 min.

All the acquired images were evaluated by a radiologist (N.Z.) with 11 years of experience in cardiovascular imaging. The radiologist was blinded to clinical findings.

Image postprocessing and evaluation were performed on a Siemens workstation (SyngoMMWP VE40A, Siemens Med Service Software, Erlangen, Germany). Stenosis and dilation of different arterial segments were evaluated by comparing the diameter of the diseased segment with the luminal diameter of the adjacent normal-caliber arterial segment. Pathological entities of each arterial segment were classified as stenosis, occlusion, and dilatation. Ten arterial segments were evaluated for each patient including bilateral subclavian arteries, brachiocephalic trunk (BT), bilateral common carotid arteries (CA), bilateral vertebral arteries (VA), ascending aorta (AA), aortic arch, and descending aorta (DA).

Image postprocessing and evaluation were performed on a Siemens workstation (SyngoMMWP VE40A, Siemens Med Service Software). The following parameters were evaluated:

• Maximum wall thickness of the thoracic aorta: The aortic wall thickness was examined at the thickest site in thoracic aorta by using dark-blood T2WI.

• Postcontrast enhancement ratio: Regions of interest (ROIs) with a size of 5 to 12 pixels were set on the thickest aortic wall in pre-, early, and delay enhanced T1WI, avoiding the adjacent intra- and extra-luminal tissue and artifact. Additional ROIs were designated on the left erector spinae muscle at the same slice. The signal intensity ratio (SIR) and enhanced SIR (ESIR) of the aortic wall to the adjacent muscle were calculated as follows.

(1)$$SIR=\frac{S{I}_A}{S{I}_M}$$

where SI${}_A$ is the SI of the thickest aortic wall on enhanced T1WI and SI${}_M$ is the SI of the left erector spinae muscle in the same slice on enhanced T1WI.

(2)$$ESIR=\frac{S{I}_A-S{I}_{Apre}}{S{I}_M-S{I}_{Mpre}}$$

where SI${}_A$ is the SI of the thickest aortic wall on enhanced T1WI; SI${}_{Apre}$ is the SI on the thickest aortic wall in pre-enhanced T1WI; SI${}_M$ is the SI on the left erector spinae muscle in the same slice on enhanced T1WI; and SI${}_{Mpre}$ is the SI on the left erector spinae muscle in the same slice on pre-enhanced T1WI.

• Aortic wall edema: The presence of aortic wall edema was considered when a high SI was noted in the thickened aortic wall on dark-blood T2WI.

The 4D flow data were analyzed using a specialized module from CVI42 software (V5.13.7, Circle, Calgary, AB, Canada) [16]. After offset correction and antialiasing during preprocessing, the entire thoracic aorta was segmented and centerline extraction was semiautomatically performed through manual modification. Subsequently, for quantitative analysis, aortic 4D flow data were acquired from the aortic root to the distal descending aorta. A semiautomatic contour detection algorithm with manual verification was used to identify the aortic plane perpendicular to the aortic centerline. PWV was automatically calculated for each patient after placing two planes at the aortic root and distal descending aorta, respectively (Fig. 3).

Fig. 3.

4D flow MRI analysis workflow. During preprocessing, the vessel area was masked for offset correction. Then, an antialiasing algorithm was used if the aliasing artifact existed in the flow region. Segmentation and centerline extraction were performed to define one mask with a threshold and the manual modification of the flow region. The centerline of the aorta was determined using the 2-point or multipoint method. Aortic plane selection: An appropriate plane of the aorta perpendicular to the centerline was selected and the contour of the aorta was manually modified. Hemodynamic analysis: Hemodynamic parameters (e.g., flow and velocity) were automatically acquired through the defined planes. PWV was automatically calculated using the phase shifting of profiles and the length between two planes along the centerline.

Statistical analysis was performed using MedCalc statistical software, version 20.019 (MedCalc Software Ltd, Ostend, Belgium). Continuous variables are expressed as the mean $\pm$ standard error or a median with 25%–75% interquartile range. Categorical variables are expressed as absolute numbers and percentages. Distribution of the normality of continuous variables was examined using the Kolmogorov–Smirnov test. Normally distributed variables were compared using the independent sample t test. Non-normally distributed variables were compared using the Mann–Whitney U test for the two groups. Categorical variables were compared using the chi-squared test. The correlation of PWV with age, disease duration, maximum wall thickness of the thoracic aorta, and delay ESIR was analyzed using Pearson’s correlation test. Receiver-operating characteristic (ROC) curve analysis was performed to compare the diagnostic performance of the maximum wall thickness of the thoracic aorta, aortic wall edema, early ESIR, the PWV of the thoracic aorta, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP). The best cutoff values were determined using Youden’s index. A p value of 0.05 was considered statistically significant.

A total of 52 patients were enrolled in this study including 23 and 29 patients with active and inactive TAK, respectively, according to ITAS2010/ITAS.A. The majority of the patients were classified as type V (34.62%, 18/52), followed by type IIb (26.92%, 14/52). In this cohort, 94.23% (49/52) of the patients were women with a mean age of 38.19 $\pm$ 12.66 years. No significant differences in sex, age, disease duration, and classification were observed between the active and inactive groups. Compared with the inactive group, the active group had higher ESR (17.83 $\pm$ 11.30 mm/h vs. 7.46 $\pm$ 4.87 mm/h, p 0.001), CRP levels (17.20 $\pm$ 27.06 mg/L vs. 0.97 $\pm$ 1.15 mg/L, p 0.001), and Takayasu clinical activity scores (ITAS2010: 6.00 vs. 0.00, p 0.001; ITAS.A: 9.00 vs. 0.00, p 0.001). Table 1 lists the clinical characteristics of the study population.

A total of 520 arterial segments were evaluated in the 52 patients. Arterial pathologies were noted in 193 segments (96 and 97 segments in the active and inactive groups, respectively), including 123 stenosed, 51 occlusive, and 19 dilated segments. The left subclavian artery (44/52, 84.62%) was the most commonly involved segment, followed by the right subclavian artery (38/52, 73.08%) and the left common carotid artery (33/52, 63.46%). Arterial dilation was more common in the aorta (12 segments) including 7 segments of the ascending aorta, 2 segments of the aortic arch, and 3 segments of the descending aorta. Table 2 presents the details of these pathologies.

The maximum wall thickness of the thoracic aorta was significantly larger in the active group (3.89 $\pm$ 1.50 mm vs. 3.00 $\pm$ 1.04 mm, p = 0.009; Table 3). Compared with the inactive group, more patients in the active group exhibited a high SI at the aortic wall (56.5% [13/23] vs. 17.2% [5/29], p = 0.003) on T2WI images, indicating the presence of aortic wall edema. In the early enhanced scan, the postcontrast enhancement ratio was significantly higher in the active group when weighted with ESIR (2.63 $\pm$ 2.14 vs. 1.11 $\pm$ 1.04, p = 0.013). Although aortic SIR and ESIR were higher in the active TAK group in delay enhanced images, no significant difference was observed between the active and inactive TAK groups. Fig. 4 presents the findings of a patient with active TAK.

Fig. 4.

MRI findings of a 32-year-old female patient with active TAK with an ITAS2010 score of 10, an ESR of 14 mm/h, and a CRP level of 32.16 mg/L. (A,B) CE-MRA showed occlusive lesions at the LSCA, RCA, and right renal artery. Significant stenosis was observed at the RSCA. (C) T2WI showed a thickened descending aortic wall with obvious edema. The thickness of the descending aortic wall was 4.65 mm. (D–F) Early and delay enhancement of the thickened descending aortic wall were observed with an early SIR of 2.20, an early ESIR of 5.46, a delay SIR of 2.95, and a delay ESIR of 6.84.

The PWV of the thoracic aorta in this cohort was 8.55 $\pm$ 4.57 m/s and significantly differed between the active and inactive TAK groups. Compared with the inactive TAK group, the active TAK group had a significantly higher PWV (10.45 $\pm$ 5.11 m/s vs. 7.20 $\pm$ 3.66 m/s, p = 0.016). A positive correlation was noted between PWV and patients’ age (r = 0.54, 95% CI = 0.27–0.73, p 0.001), disease duration (r = 0.52, 95% CI = 0.25–0.71, p 0.001), maximum wall thickness of the thoracic aorta (r = 0.45, 95% CI = 0.19–0.65, p = 0.001). However, a negative correlation was noted between the PWV and delay enhanced ESIR (r = –0.33, 95% CI = –0.59 to –0.0068, p = 0.046; Fig. 5).

Fig. 5.

Correlation between PWV and different parameters. (A–C) A positive correlation existed between PWV and patient’s age (r = 0.54, 95% CI = 0.27–0.73, p 0.001), disease duration (r = 0.52, 95% CI = 0.25–0.71, p 0.001), maximum wall thickness of the thoracic aorta (r = 0.45, 95% CI = 0.19–0.65, p = 0.001). (D) A negative correlation existed between the PWV and delay enhanced ESIR (r = –0.33, 95% CI = –0.59 to –0.0068, p = 0.046).

Among all the MRI indicators of the thoracic aorta, the area under the curve (AUC) was the largest for the maximal wall thickness (0.804, 95% CI = 0.667–0.941). According to the ROC curve analysis of these MRI indicators, the cutoff values of the maximal wall thickness, early enhanced ESIR, and PWV were 3.12 mm, 1.13, and 6.4 m/s, respectively. On the basis of these cutoff values, the maximal wall thickness exhibited the highest sensitivity for the determination of active TAK (93.33%, 95% CI = 68.1%–99.8%), whereas wall edema demonstrated the highest specificity (84.00%, 95% CI = 63.9%–95.5%; Table 4).

ESR (0.851, 95% CI = 0.733–0.969) and CRP levels (0.808, 95% CI = 0.647–0.969) were superior to the MRI indicators of the thoracic aorta for determining the clinical activity of TAK (Fig. 6).

Fig. 6.

AUC determined through the ROC analysis of MRI indicators. ESR (0.851, 95% CI = 0.733–0.969) and CRP (0.808, 95% CI = 0.647–0.969) were superior to each thoracic aortic wall characteristic for the assessment of TAK activity. Among all MRI findings, AUC was the largest for the maximal wall thickness (0.804, 95% CI = 0.667–0.941).