Giant cell arteritis (GCA), or temporal arteritis, is the most common form of systemic inflammatory vasculitis in adults. GCA is a panarteritis that insults older people almost exclusively. Because extracranial branches of the carotid artery are commonly affected, the most dreadful complication of the disease is blindness, which should be prevented by high index of suspicion and timely treatment with corticosteroids [1]. Given the lack of specificity both of the clinical and laboratory data [2], a temporal artery biopsy is always included in the algorithm for the approach to diagnose GCA [3]. For purposes of classification, the American College of Rheumatology (ACR) proposes that a patient with vasculitis is said to have GCA if three of the following five criteria are present: (a) age 50 years or above, (b) new onset of localized headache, (c) temporal artery tenderness on palpation or decreased pulsation, (d) erythrocyte sedimentation rate (ESR) of 50 mm/hour or more, and (e) histologic findings [4]. However, although a temporal artery biopsy is the diagnostic gold standard for this disease [5–7], biopsy results may be negative in 9% to 44% of patients with a clinical diagnosis of GCA [6–8].

Schmidt and colleagues were the first to propose ultrasonography as a quick, easy, and non-invasive test for identifying GCA, before biopsy [9–11]. These studies suggested that the presence of the halo sign (that is, a dark area around the vessel lumen probably due to arterial wall edema) is highly specific for GCA. This hypoechoic finding is circumferential and has to be demonstrated in two planes. Apart from the halo sign, stenoses and occlusions of the temporal arteries, frequently detected in acute disease, were recognized by Schimdt and co-authors as additional characteristic signs in patients suspected of having GCA [9, 10]. The diagnostic value of colour duplex sonography (CDS), however, remains the subject of debate; Salvarani et al. have found that CDS does not improve the diagnostic accuracy of a careful clinical examination [12]. To further examine the diagnostic value of CDS in patients with suspected GCA, we designed a prospective study to specifically test the hypothesis that CDS can replace biopsy in the algorithm for the approach to diagnose GCA in the clinical setting.

A careful physical examination, including palpation of the temporal arteries, accompanied by an accurate medical history and laboratory data are all imperative for the diagnosis of GCA [12, 16]. Moreover, clinicians should be familiar with the atypical presentation of the disease, including fever of unknown origin, respiratory tract symptoms, and large artery involvement [1]. Symptoms such as jaw claudication and diplopia significantly increase the probability to diagnose GCA [16, 17]. In our prospective study, visual disorders and jaw claudication at baseline were associated with high specificity, albeit with low sensitivity. Although jaw claudication had 100% specificity, only three of our 22 patients with GCA reported this symptom. This frequency is clearly lower than expected [1–3], but similar findings have been reported in other cohorts of patients with GCA [18].

Accumulated evidence suggests that of greater sensitivity than clinical evaluation of patients are findings demonstrated by CDS of the temporal arteries [9–11, 18, 19]; CDS is of additive value in assessing the peripheral involvement of the disease [20] or in diagnosing concomitant GCA in patients with polymyalgia rheumatica [21]. Despite the fact that temporal artery biopsy is considered the gold standard for the diagnosis of GCA, patients with definite GCA frequently may have a negative biopsy [2, 6, 8]. In addition, the need for a large arterial segment (2 to 3 cm) due to the segmental nature of the vascular inflammation [22–25] and the practical difficulties in repeating it, make temporal artery biopsy not a simple procedure [26]. The present study was designed to specifically examine the diagnostic value of CDS with respect to common problems in the approach of a patient with clinically suspected GCA, for whom the decision for immediate treatment is imperative.

In agreement with previous studies [18, 19], we found that the ultrasonographic evidence of a halo around the temporal artery lumen yields a fair sensitivity (82%) and a high specificity (91%) for the diagnosis of GCA. Although Salvarani et al. reported that only halos greater than 1 mm in thickness increased the probability of a positive biopsy result [15], our findings suggest that even small halos of 0.7 mm may predict GCA. Moreover, bilateral halos were evident in almost one third of patients with GCA, and the specificity of this finding for GCA diagnosis reached 100%, suggesting that in a given patient with clinically suspected GCA and bilateral halo signs, a temporal artery biopsy is not necessary. To the best of our knowledge, the specificity of this particular sign in CDS has not been examined previously.

It is well known that CDS is an investigator-dependent test in which the skill and the experience of the operator are important variables in determining the diagnostic accuracy of the method. For example, if colour setting is not properly selected, the arterial representation is not optimal and a thin halo may be missed in case colour covers both the artery wall and the lumen (that is, 'colour bleeding', false-negative finding). On the other hand, if colour covers only the centre of the artery, the periphery of the artery appears dark and may be misdiagnosed as a halo (false-positive finding). A formal evaluation of inter-reader variability was not conducted in our study; such data could help in minimizing false-positive or -negative findings which can be obtained even by experienced sonographers. Along this line, although a true positive finding cannot be ruled out in our patient with tuberculosis, the halo that was demonstrated in this particular case may represent a low examination quality. Moreover, the examination should be performed by a modern scanner with adequate resolution to investigate the small temporal arteries placed near the skin surface [10]. In agreement with previous reports [9, 27], halos disappeared within a mean of 22 days after the initiation of corticosteroid treatment in our patients. The true interval for halo disappearance could be even shorter if closer follow-up points had been chosen. It should be noted, however, that directional biopsies performed in patients with unilateral halos may have partly influenced this result. Finally, we noticed that in four patients in whom GCA relapsed during tapering of corticosteroid treatment, halos reappeared and regressed again when the corticosteroid doses increased. Thus, although not systematically studied, CDS may be of significant value also in cases of GCA relapses.

We also found that directional temporal artery biopsies in all patients with halos and GCA were always positive, indicating that CDS examination before performing a biopsy could avoid 'generous' excisions of the temporal artery in many cases. Previous studies comparing CDS and histology of the temporal artery in GCA suggested that patients with halos had a more pronounced inflammatory cell infiltration in biopsy whereas patients without halos demonstrated histological signs of subtle inflammation [28]. A formal evaluation of the diagnostic value of the temporal artery biopsy alone, as well as a comparative analysis of CDS and biopsy findings, is not possible given that 20 of 49 biopsies were directed and 18 of them were positive for GCA, thus biasing any comparison. Regarding the presence of stenoses and occlusions in patients with GCA, the results reported here are in contrast to those of Schmidt and colleagues [9]. Blood-flow abnormalities were common in GCA and non-GCA patients, as well as in elderly healthy subjects and patients with macrovascular disease associated with diabetes mellitus or stroke (Table 2), suggesting that the presence of stenoses and occlusions of the temporal arteries does not increase the diagnostic value of CDS for GCA, at least in our hands. In a recent meta-analysis examining the results of 23 relevant studies, it was concluded that although CDS is a relatively accurate diagnostic method, cautious interpretation of the test results in terms of clinical manifestations and pretest likelihood is essential [29]. However, these conclusions are based on the combined analysis of any kind of abnormality found in CDS in patients with clinically suspected GCA, including perfusion and blood-flow abnormalities, and the diagnostic value of bilateral halo signs was not examined [29]. Moreover, all relevant studies have used the ACR classification criteria for GCA [4] as the reference standard for CDS. Such comparisons may be biased because these criteria have been developed for patients with vasculitis, whereas in the clinical setting a given elderly patient with suspected GCA may suffer from infectious or neoplastic diseases, as was the case in up to one third of the patients in our cohort.

Finally, our results may allow us to partially modify an established algorithm for the approach to diagnosing GCA, introduced by Hellmann and Hunder [3]. As depicted in Figure 5, we propose that after a careful clinical examination and assessment of relevant laboratory data, CDS examination should precede the biopsy in patients with suspected GCA. Among the various abnormalities that can be found in CDS, only the halo sign should be considered. In case of bilateral halo signs, treatment should be initiated without proceeding with biopsy. If unilateral halos are present, a decision of directional biopsy is justified. If the histology result is negative, clinicians should follow the algorithm introduced by Hellman and Hunder [3].

CDS examination is an inexpensive, non-invasive, reproducible, and easy-to-perform method that should precede temporal artery biopsy in all patients with suspected GCA. Biopsy is not necessary in a substantial proportion of patients in whom bilateral halo signs around the temporal arterial wall are found in CDS. The presence of a unilateral halo allows a directional biopsy that has an increased probability to confirm the clinical diagnosis. Further studies to assess the test performance of CDS in assessing relapses of GCA are warranted.