Takayasu arteritis (TA) is a chronic, idiopathic inflammatory disease that principally affects the aorta and its primary branches (1, 2). TA typically occurs in women, most often during their reproductive years. Men are a minority in all series. Vessel wall inflammation may cause organ ischemia due to arterial stenosis or occlusion. Less often, aneurysms may produce sequelae such as aortic regurgitation and, rarely, vascular dissection or rupture. It is therefore crucial that active disease be treated before vascular inflammation produces anatomic abnormalities.
It has often been assumed that active vasculitis of any type is usually clinically apparent and correlates with abnormal levels of acute-phase reactants. However, this observation has not withstood rigorous evaluation, especially in regard to TA. In TA, clinical signs and symptoms and measures of acute-phase reactants may not be sufficiently sensitive to identify ∼50% of all patients with active disease (1–4). These observations are based on identification of active vasculitis in >40% of specimens obtained during bypass surgery from patients whose TA was thought to be quiescent, as well as new angiographic abnormalities at previously unaffected sites in 61% of patients whose TA was thought to be in remission (1, 2).
Past attempts to directly visualize large vessel inflammation by indium-111 leukocyte scanning have been shown to be relatively insensitive (5). Basic “edema-weighted” magnetic resonance (MR) imaging has been shown to be highly sensitive in demonstrating areas of extracellular fluid, which manifest as increased signal intensity compared with surrounding normal tissue. These sequences have been used to identify tissue edema, with or without inflammation, in patients with musculoskeletal diseases (6), tumors (7), liver disease (8) and vascular graft infections (9). Recent methodologic developments have permitted tissue characterization, including detection of edema, to be performed rapidly and relatively free of artifacts (10, 11).
Limitations in the assessment of TA activity led to our prospective evaluation of the utility of MR in 24 TA patients. The goal of this study was to assess the performance characteristics of advanced “edema-weighted” MR techniques in imaging large vessel topography and recognizing qualitative features of vessel wall injury that reflect inflammation.
- Top of page
- PATIENTS AND METHODS
MR findings in the TA patients. There were 5 males and 19 females in the study population. Their ages ranged from 14 to 57 years (mean 34 years). Nineteen (79%) were Caucasian, 3 were Middle Eastern, 1 was Hispanic, and 1 was Indian. Seventy-seven MR examinations were performed on these 24 patients between November 1, 1995 and April 20, 1999 (Table 3). Sequential imaging was performed on 16 patients. MR readings for each examination were performed by the 2 collaborating cardiovascular radiologists, using the aforementioned rating scale.
Table 3. Association between acute-phase reactants and MRI findings in patients with Takayasu arteritis
| ||Clinical diagnosis*|
|Active (n = 18)||Inactive (n = 43)||Uncertain (n = 16)||Inactive + uncertain (n = 59)|
|% with abnormal ESR (normal ≤30 mm/hour)||57||29||36||31|
|% with abnormal CRP (normal ≤2 mg/dl)||29||29||14||26|
|ESR, mm/hour (normal ≤30 mm/hour)|
|Enhancement on MR|
| % of patients||94||56||81||63|
| % with abnormal ESR or CRP||62||29||23||30|
|Vessel wall thickening, mm (normal 1–2 mm)|
|Vessel wall edema rating (0–3 scale)‡|
The frequencies of new or progressive features of active disease are presented in Figure 1. Figures 2–4 present examples of vessel wall thickening (Figure 2) and edema (Figures 3 and 4) and response to therapy (Figure 3) as demonstrated by MR.
Figure 1. Frequency (%) of new-onset or progressive symptoms during disease exacerbations in patients with Takayasu arteritis. BP = blood pressure; CNS = central nervous system.
Download figure to PowerPoint
Figure 2. Increased aortic wall thickness in Takayasu arteritis, as demonstrated on magnetic resonance images. Transaxial (top) and longitudinal-oblique (bottom) images of the thoracic aorta reveal abnormal vessel wall thickening (arrow). Lobular ectasia is visible in the ascending and descending thoracic aorta, the aortic arch, and the innominate and left subclavian arteries.
Download figure to PowerPoint
Figure 3. Response of aortitis to effective therapy, as demonstrated on magnetic resonance images. At initial presentation, “edema-weighted” images (top) revealed circumferential wall thickening (5 mm), with moderately increased signal intensity of the thickened wall relative to skeletal muscle (arrow). Following aggressive immunosuppressive therapy (bottom), corresponding images showed improvement in wall thickening and reduced signal intensity. These changes are compatible with diminished inflammation. T2-weighted images are shown on the left; STIR images are shown on the right.
Download figure to PowerPoint
Figure 4. A, Magnetic resonance images of the aorta of a 35-year-old woman with shortness of breath and angina. The patient underwent surgery for an ascending aortic aneurysm and valvular regurgitation. Images obtained 2 days before surgery showed lobular dilatation of the aorta, increased wall thickening, and increased signal intensity from the aortic root to the level of the diaphragm. Increased wall signal intensity was consistent with active inflammation. STIR = short tau inversion recovery. B, Histopathologic section of the aorta, showing transmural mononuclear cell inflammation, which includes giant cells (hematoxylin and eosin stained).
Download figure to PowerPoint
Among the 18 patients with clinically unequivocal active disease (Table 3), concurrent MR images revealed increased signal intensity consistent with edema in all but 1 case (94%). The enhancement rating in these patients ranged from 0 to 3 (mean 2.2). Only 62% of these patients had concurrently elevated erythrocyte sedimentation rate (ESR) or C-reactive protein (CRP) values.
In 16 instances, MR studies were performed when disease activity was clinically uncertain. In 13 of these 16 studies (81%), vessel wall edema was present. The enhancement rating in these patients ranged from 0 to 3 (mean 1.9). In only 3 of this subset of 13 patients with vascular edema noted on MR was there an increase in levels of acute-phase reactants (23%).
Twenty-four of 43 MR studies obtained during periods of clinically presumed remission demonstrated increased signal intensity consistent with edema (56%). The enhancement rating in these patients ranged from 0 to 3 (mean 1.2). In only 29% of these patients in whom MR results were suggestive of active vasculitis (i.e., vascular edema present) did patients have abnormal ESR or CRP values.
We found no statistically significant associations between clinical characteristics or acute-phase reactant levels and edema-weighted features of vascular enhancement on MR. There was, however, the suggestion of a trend toward higher vessel wall edema on MR (P = 0.07) and abnormal levels of acute-phase reactants (ESR or CRP; P = 0.06) in the group with unequivocal clinically active disease compared with the group with inactive disease.
Sixteen patients underwent several MR studies over time. A total of 12 new anatomic abnormalities (4 occlusions, 7 stenoses, and 1 dilatation) were demonstrated over time in 8 of these patients. The range of MR vessel edema scores concurrent with these new findings was 0–3 (mean 1.25). Among patients in whom sequential MR studies were repeated over at least 5 months, 2 patients did not have vessel edema and did not develop new lesions, 6 did not have disease progression despite the presence of vessel edema on consecutive MR studies, 3 patients developed 6 new lesions in the absence of concurrent edema, and 5 had new lesions after the appearance of vessel edema. Concurrent therapy may have played an important role in modifying these results. Only 3 of the 16 patients were not receiving immunosuppressive therapy at the time the new lesions were demonstrated on MR. One of these 3 untreated patients, who appeared to be in remission, had anatomic progression but did not have vessel edema on sequential studies.
Two of our 24 patients required surgery at the time that MR was being used to evaluate large-vessel inflammation. Both were women (ages 35 and 44) who had severe aortic regurgitation and root dilatation. Neither was systemically or chronically ill or known to have active inflammatory vascular disease before the MR images were obtained. In both cases, MR revealed 3–4 mm of thickening and increased signal intensity of the ascending aorta and aortic arch. The enhancement rating in these patients was 1 and 3, respectively. One patient had wall thickening extending to the descending aorta, with lobular dilatation of the entire thoracic aorta. The other patient had remarkable thickening and signs of edema extending to the proximal arch vessels and pulmonary arteries. The histopathologic specimens from both patients revealed active granulomatous inflammation (Figure 4).
Results of MR interobserver agreement analysis. For one MR interpretation agreement study, random image sampling from 59 visits was employed. In these 26 patients, T2-weighted and STIR images were evaluated by both readers. Results are given below, first for the ordinal outcome (a reading of 0–3) and then for the binary outcome (0 versus 1–3).
Ordinal outcome. Using the Fleiss and Cohen weighting system, the agreement beyond chance (i.e., kappa statistic) was 0.82 (95% confidence interval [95% CI] 0.66–0.99) for T2 and 0.84 (95% CI 0.68–0.92) for STIR readings. These both indicate very good to excellent agreement beyond what would be expected just by chance. Using the Conchetti and Allison weighting scheme, which gives smaller weight to disagreements, the kappa statistics were 0.76 (95% CI 0.55–0.96) and 0.78 (95% CI 0.63–0.86) for T2 and STIR readings, respectively, which is also very good agreement. We found no statistical difference in agreement between the T2 and STIR methods for either the random or followup studies.
Binary outcome. Between-reader agreement was also determined in regard to their assessment of images as showing either positive or negative (reading of >0 or 0, respectively) enhancement. Kappa statistics were 0.84 (95% CI 0.66–1.0) and 0.78 (95% CI 0.56–0.95) for random T2-weighted and STIR readings, and 0.89 (95% CI 0.78–1.0) and 0.83 (95% CI 0.55–1.0) for followup T2-weighted and STIR readings, respectively. Again, no statistical differences were found. For binary outcome analyses, there were few disagreements: only 4 of 59 and 5 of 59 for T2 and STIR readings, respectively, for randomly presented studies, and 2 of 46 and 3 of 46 for T2 and STIR readings, respectively, among studies presented in temporal sequence of acquisition.
We concluded that agreement between the 2 readers was good to excellent, using either the T2-weighted or STIR methods and that neither method stands out as being better than the other in the evaluation of vessel enhancement. These observations provided confidence in the consistency of interpretations of the MR studies obtained in our clinical study group. We used the evaluations of either reader in comparing the clinical status and MR results as shown in Table 3.
- Top of page
- PATIENTS AND METHODS
This study is the first prospective analysis of the utility of recent technical advances in MR imaging in patients with Takayasu arteritis. We have demonstrated that this radiation-free, noninvasive means of evaluating large vessels can be an important supplement to following anatomic changes in TA patients. By incorporating edema-weighted images, we had hoped to demonstrate that MR may also be a useful adjunct to clinical and laboratory assessment of disease activity. Patients in our study with unequivocal clinically active TA had a high frequency of vessel wall edema (94% of studies), as determined by increased MR signal. Signs of vessel wall edema by MR were also demonstrated in >50% of all patients in whom TA was judged to be clinically inactive or of uncertain activity status.
Because most of the patients in this series did not require surgery, questions remain about whether the findings of vessel edema include numerous false positives, a product of excessive MR sensitivity. The credibility of the data is supported by independent observations. First, histopathologic evidence of active vasculitis has been noted in 42–44% of vascular bypass specimens from patients who were clinically assessed to be in remission at the time of bypass surgery (1, 2, 4). Second, in a National Institutes of Health report of 60 patients, sequential angiographic studies demonstrated new lesions in 61% of those who were thought to be in remission (1, 2). This proportion of TA cases in whom asymptomatic active arteritis was present is similar to that in our MR study. Nonetheless, these observations offer only indirect, circumstantial evidence that the finding of edema-enhanced vessels on MR represents inflammation. It is theoretically possible that the vessel wall could remain edematous for unknown periods of time after effective treatment of inflammatory lesions. The lack of a consistent correlation between edema and subsequent changes in vascular anatomy introduces further uncertainty about the utility of MR-determined vessel edema as a measure of disease activity.
Since the completion of this study, we have also evaluated vascular MR studies in 34 additional patients who have required aortic root reconstruction, with or without valve replacement (Hoffman G, et al: unpublished observations). In 10 instances, histopathologic findings and edema-weighted MR images could be compared. Previously recognized giant cell arteritis (GCA) was present in 4 patients. Three patients had histologic proof of GCA discovered from their surgical specimens. One patient had relapsing polychondritis and a thoracic aortic aneurysm. Two patients had atherosclerosis. In 4 patients (1 with GCA and 3 with idiopathic aortitis discovered in surgical specimens), the findings of aortic wall edema correlated with histopathologic evidence of aortitis. In 3 patients (2 with atherosclerosis and 1 with relapsing polychondritis in remission), the absence of MR edema correlated with an absence of histopathologic evidence of inflammation. However, in 3 patients (known GCA), vascular edema was present on MR, but histologic evidence of inflammation was absent. Although these 3 cases would appear to be false positives for edema/MR-identified inflammation, this 30% false-positive rate may be related to tissue sampling in a disease that is known for “skip lesions.”
This study differs from other studies that describe wall thickening and luminal changes with MR or CT scanning (16–25) or thrombus formation on vessel walls (25). In TA, luminal abnormalities and wall thickening alone cannot distinguish active inflammation from chronic fibrotic lesions (26). A previous report of MR studies in TA emphasized abnormalities of the vessel wall, based on contrast-enhanced imaging, which required use of an intravenous iodinated agent. That study also emphasized the weak correlation of ESR or CRP values with vessel “inflammation” (edema), as indicated by tissue enhancement on MR (27). Unlike that study, our investigation utilized a form of MR-inherent signal changes that are related to changes in the distribution of tissue water due to edema, rather than differences in the distribution of extracellular contrast agent. The technique utilized in our study does not require contrast agents.
We are not suggesting that vascular MR replace invasive angiography. Each technique has unique advantages and disadvantages (Table 4). Sequential invasive angiography has been very informative in detecting new vascular lesions in TA patients (1, 2). It provides greater image resolution than MR in vessels smaller than the aorta, such as the innominate, renal, mesenteric, and iliac arteries (28), which makes it desirable for the preoperative planning of bypass surgery. Invasive angiography provides opportunities to perform transluminal angioplasty, and it enables the recording of central blood pressure readings and gradient determinations in patients with vessel stenoses, which allows the clinician to assess the reliability of extremity blood pressure cuff measurements. Alternatively, the noninvasive, radiation-sparing qualities of MR with “edema-weighted” imaging techniques may lead to its preferred use for routine followup.
Table 4. Properties of “edema-weighted” magnetic resonance imaging and invasive agniography
|Magnetic resonance imaging||Angiography|
| Noninvasive||Better image resolution|
| Defines anatomy||Ability to perform interventions|
| Can evaluate wall thickness Can determine enhancement||Can record intravascular blood pressure and measure gradients|
| No radiation exposure|
| Resolution lower than that of angiography||Invasive|
| Not quantitative in assessment of signal intensity||Evaluates lumen (“lumenogram”)|
| Does not provide intravascular pressure||Ionizing radiation|
|Impractical for frequent monitoring|
Our experience lends further support to past impressions that clinical features of disease activity and levels of acute-phase reactants underestimate active, progressive TA in at least 50% of patients. We have found MRI to be an important adjunct to the assessment of TA. “Edema-weighted” images detect fluid within the vessel wall, which may be due to inflammation. However, it is possible that edema may remain during tissue remodeling, after inflammation has resolved. The finding of new anatomic lesions in conjunction with “edema” would favor the presence of active disease. However, the absence of a consistent correlation between MR findings and clinical and laboratory parameters of disease activity, as well as the development of new lesions, has influenced us to not use MR-determined vessel edema as the sole guide to disease activity and treatment. Additional studies that provide MR images and concurrent histopathologic specimens will be required to determine whether MR vessel “edema” correlates with inflammation and whether treatment based on such findings alters disease prognosis.