Historically, the diagnosis of GCA has been made on clinical grounds and confirmed with temporal artery biopsy. There has been a shift in recent years to using US and MRI as first-line investigations for suspected cranial LVV and is a new recommendation by the EULAR 2018 guidelines for imaging in LVV [12, 20, 21]. A recent meta-analysis reported US of the temporal arteries had a pooled sensitivity of 77% and specificity of 96%. High-resolution 3-T MRI had a pooled sensitivity of 77% and specificity of 88% [21]. The EULAR recommendations for imaging in LVV state a need for prospective studies directly comparing US and MRI.
The EULAR recommendations for imaging in LVV recommend the use of high-resolution 3-T MRI; however, studies comparing high-resolution 3-T MRI to US may not be applicable in practice as most centers only have access to 1-T or 1.5-T MRI machines [12, 21].
Our study showed that US was able to detect vasculitic changes similarly to 1.5-T MRI. There was no statistical difference between US and MRI in any of the cranial vessels, except the right frontal artery where MRI detected vasculitis in 17 (52%) patients and US detected changes in 7 (21%) (p < 0.002). Previous studies suggest 1.5-T MRI is less sensitive than US for detecting temporal artery vasculitis; however, our study did not support this finding [22].
This is the first study to compare US to MRA for the examination of extracranial large vessels in GCA. US detected vasculitic changes in the supra-aortic arteries more frequently than MRA. For example, US detected disease in the left axillary artery in 19 patients (58%) and in the right axillary artery in 25 patients (76%) compared to 3 (9%) and 5 (15%) with MRA, respectively (p < 0.001 and < 0.0003, respectively). This may be due to the fact that US provides higher resolution than contrast-enhanced MRA. MRA provides information about the lumen of the artery, and inflammation is assessed indirectly by stenosis. Minor vasculitic changes within the walls of the vessel may not have been detected without additional transverse post-gadolinium sequences in the large vessels. The abdominal aorta was omitted from analysis because additional MRA images would be required to evaluate the vessel in full. This may have contributed to the lower sensitivity of MRA compared to US in the large vessels. Because the axillary artery is one of the most common vessels involved in GCA, these findings suggest that US will perform better than MRA for the detection of GCA in the large arteries.
In patients with new-onset disease, glucocorticoids were initiated at the time of US examination, but for those with chronic disease, data regarding the initiation of glucocorticoids was not obtained. One study found that the sensitivity of US and MRI to detect vasculitis in the cranial vessels decreases similarly with glucocorticoid use [23]. There is marked loss of sensitivity of US to 50% and MRI to 56% after more than 4 days of glucocorticoid treatment. However, there are no studies comparing the effects of glucocorticoid use on US and MRA of the large vessels. The use of glucocorticoids may affect the sensitivity of MRA more so than US and may account for the higher detection rate with US. Furthermore, this may explain the higher concordance seen between MRA and disease activity. Resolution of contrast enhancement on MRA is reduced after 4–5 days of glucocorticoid therapy, whereas large vessel changes on US may remain detectable for months in the majority of patients, even after normalization of systemic inflammatory markers [24, 25]. The probability for MRA to detect vessel wall enhancement after more than 5 days of glucocorticoid therapy is reduced by 89.3% [26]. The lasting changes seen with US may make US the preferable imaging modality for disease detection.
US is able to detect vasculitic changes in the cranial and extracranial arteries, seen as a halo sign, a positive compression sign, stenosis, or occlusion. When extracranial arteries are examined, there is a marginal improvement in sensitivity by 2%, as seen in a study that compared examination of the temporal and axillary arteries compared to temporal artery alone [27]. However, in clinical practice, most protocols include the examination of additional cranial and large vessels. In our study, US detected vasculitic changes more frequently in the extracranial arteries than MRA, but the study was not designed to examine whether evaluating additional arteries increased the sensitivity of US for the diagnosis of GCA. Further research is needed to clarify whether the evaluation of additional vessels increases diagnostic accuracy.
Rapid diagnosis of GCA using US has resulted in a significantly reduced number of GCA patients with vision loss [28]. With US, the radiation exposure of CTA and FDG-PET/CT is avoided. Furthermore, patients who have contraindications to the use of MRI and contrast agents can be imaged with US. US is less time consuming and cheaper compared to MRI. Taking into account that US detects more abnormalities in the large vessels than MRA, and the reduced costs to perform the examination compared to MRI, one could conclude that US is more cost-effective than MRI. However, our study was not designed to examine the cost-effectiveness of the two imaging modalities and this conclusion should be interpreted with caution.
The excellent interobserver agreement among the sonographers and the moderate agreement among the radiologists could be attributed to the different experience among the readers. While both ultrasonographers performed vascular ultrasound for over 5 years, the local radiologist (FL) had limited experience on the interpretation of MRI and MRA images. This finding underscores the importance of adequate training programs for all personal performing or interpreting imaging modalities which are visualizing vascular structures.
The limitations of this study include the small sample size (n = 35) and cross-sectional study design, whereby treating physicians were not blinded to the patient’s clinical presentation when interpreting imaging results. This was controlled for by having external blinded reviewers interpret the images. Of note, there was no control group presented to radiologists for comparison.
Data in this study was collected before cut-off values for normal IMC thickness had been defined. Abnormal IMC thickness was later defined as 0.42, 0.34, 0.29, 0.37, and 1.0 mm for the common temporal, frontal temporal, parietal temporal, facial, and axillary arteries, respectively. In addition, compression testing on US was not performed in this study. Both could have had an impact on the accuracy of US.