Introduction

Giant cell arteritis is the commonest form of vasculitis seen in Caucasian population and is a medical emergency due to risk of acute visual loss or stroke [1]. Temporal artery biopsy (TAB) has traditionally been considered the gold standard investigation to diagnose giant cell arteritis (GCA), if performed within the right time frame. The British Society of Rheumatology (BSR) guidelines from 2010 recommend that a surgical biopsy with a minimum of 1 cm of temporal artery is obtained for each patient and the procedure should ideally be performed within 1 week of starting corticosteroids, by a trained surgeon with experience in the technique [2]. Temporal artery Ultrasound (US) has taken on a more leading role in the last decade in fast track pathways, as characteristic changes such as “halo” can be seen, enabling early non-invasive confirmation of diagnosis, reducing exposure to corticosteroids for patients who do not have GCA and also improving visual outcomes [3]. Also, this can be repeated for flares. US has gained popularity in fast track pathways (FTP) over the last decade as it is easy to obtain, cheap and non-invasive.

Studies have compared the sensitivity and specificity of US with TAB. The TABUL trial investigated the role of US compared with TAB in the diagnosis of GCA, where 381 patients underwent both ultrasound and TAB. US demonstrated higher sensitivity compared to TAB (54% vs 39%), but a lower specificity (81% vs 100%). The authors suggested sequential investigation for GCA by performing US first, followed by TAB in US negative cases, which increased the sensitivity to 65%, with specificity of 81% and reduced the need for biopsies by almost 43%. However, within the TABUL trial, the authors also reported that 56% of TAB specimens did not meet BSR standards with 43% of the TAB arterial specimen length being less than the BSR recommended length of 1 cm and 13% of specimens were not arteries resulting in a failed biopsy. This could be due to practical difficulties in performing TABs by surgeons with adequate expertise. These factors are likely to have contributed to the lower diagnostic yield and therefore a relatively lower sensitivity of TAB within this study [1]. Other studies also recommend US as the first line of investigation for GCA, and TAB should be performed on those with negative US results [4,5,6,7] since there appears to be good correlation between halo and positive TAB [5].

Temporal artery biopsies have become the second line of investigation in fast track pathways that use US first, but it is unclear how useful TAB is in this setting due to paucity of data on the role of TAB in patients with negative US. This has remained an unmet need for appropriate management of GCA. There also remains an ongoing need to assess impact of arterial specimen length and duration of treatment with corticosteroids on sensitivity of TAB, as there is limited evidence in this area within a GCA FTP.

A literature search including PubMed, Embase, Clinical Key incorporating Medline, and Google Scholar returned no reports in literature that have prospectively evaluated the role of TAB in a sequential GCA FTP. The Coventry sequential GCA FTP was established in 2013 and runs as a shared service amongst departments of Rheumatology, Ophthalmology and Vascular physiology [8]. The aim of this study is to evaluate the role of TAB in this sequential GCA FTP, especially looking at its utility in those with negative or inconclusive US. The study also looked at how the arterial specimen length and duration of treatment with corticosteroids prior to the biopsy had an impact on its sensitivity.

Materials and methods

A prospective database was maintained for all patients who presented to University Hospital Coventry and Warwickshire NHS Trust (UHCW) with symptoms of GCA and referred for temporal artery biopsies from May 2014 to June 2019 via the GCA FTP. Ethical approval was obtained from the University Hospitals of Coventry and Warwickshire Research and Development department (GF 0264).

Patients were assessed and managed under the Coventry sequential GCA FTP [8]. UHCW established a novel multi-disciplinary sequential GCA FTP in 2013, to facilitate a streamlined process for the management of GCA. Patients are assessed on the same day and if suspected to have GCA, they have blood investigations and are referred for a walk-in vascular Doppler temporal artery US. US was usually performed on the day of the referral or next day if presented out of hours.

The US was performed by vascular scientists who have undergone formal accreditation through the Society for Vascular Ultrasound Accreditation Scheme. In addition, specific training around temporal artery US for diagnosis of Giant cell arteritis, was performed in-house. This in-house training is in line with the subsequently published document outlining training requirements [9]. The images were audited by the lead vascular scientist in 2017 and no major discrepancies were found.

If US is positive and supported by the clinical and laboratory findings, then the patients are treated for GCA. A positive US was determined by the presence of a halo, which was considered a definitive sign for GCA. Other findings such as thickening or stenosis were considered inconclusive on US. Patients with low clinical suspicion and negative US are discharged on the same day with no steroid treatment. If US is negative or inconclusive, but there is a moderate-to-high clinical suspicion for GCA, then patients are started on corticosteroids and referred to the named Ophthalmology consultant (PM) for a TAB, which is usually performed within 2 weeks from initiating corticosteroids. The criteria for a positive biopsy findings for GCA included lymphocytic and macrophage infiltration of the arterial wall + /− presence of multinucleated giant cells, fragmentation of the internal elastic lamina and intimal hyperplasia.

Data collection included patient demographics, clinical features, blood investigation results, histological findings including post-fixation arterial specimen length, ultrasound findings, diagnosis, and treatment including duration of corticosteroid use. All patients with failed biopsy or where specimen length did not meet BSR guidelines of at least 10 mm were excluded from the final analysis.

Data were analysed to assess sensitivity and specificity of TAB in all patients where arterial specimen met BSR guidelines. We compared the sensitivity and specificity of TAB in patients with a reference clinical diagnosis of GCA made by the treating physician based on clinical features and available test results. Further analysis looked at the sensitivity and specificity of TAB in those patients who had a negative or inconclusive US.

We analysed how the post-fixation arterial specimen length affected the sensitivity of TAB in 5 mm increments (10–14 mm, 15–19 mm, 20–24 mm, 25–30 mm, >30 mm). We looked at how duration of treatment with corticosteroids (typically 60 mg oral Prednisolone if jaw claudication or visual symptoms, 40 mg if no worrying features) prior to biopsy affected the sensitivity of TAB. This was subdivided by steroid duration in 10-day increments (i.e., 0–10 days, 11–20 days, >20 days) and weekly increments (0–7 days, 8–14 days, 15–21 days, >21 days). We evaluated the patients clinical and laboratory characteristics to assess whether any of these, increase the odds of having a positive biopsy, using univariate logistic regression analysis.

Statistical analysis was performed using the “R” statistical package. P < 0.05 were considered statistically significant.

Results

A total of 1149 patients with suspected GCA, were referred to the Coventry sequential GCA FTP, between May 2014 and June 2019. During this 5-year period, 109 patients (9.5%) were referred for TAB, via the Coventry sequential GCA FTP.

Exclusions: Five out of the 109 patients who had TAB, were excluded from the analysis, as the arterial specimen did not meet the BSR standards. Three had an arterial specimen length <10 mm, one arterial specimen length was not reported and one specimen was a vein.

The median age of the patients was 72 years (mean: 71.8, range 51–95), with 67% (70) being females. Headache (72%), visual symptoms (54%) and scalp tenderness (47%) were the commonest presenting features. Table 1 demonstrates the clinical features and blood investigation findings for all patients referred for TAB and for those patients diagnosed with GCA.

Table 1 Clinical and laboratory characteristics in patients referred for TAB and in those diagnosed with GCA.
Full size table

Sensitivity and specificity of TAB in patients when arterial specimen met the BSR guidelines

We analysed 104 patients, of whom 57 (55%) were given a final diagnosis and treatment for GCA. Amongst these 57 patients, TAB was positive in 27 and negative in 30.

In the remaining 47 patients who were biopsy negative, treatment for GCA was discontinued. Other pathological diagnoses included arteriosclerosis in 7 patients and another form of vasculitis in one. The median length of biopsies in this cohort was 20.5 mm, (mean: 20.5 mm) with a range of 10–36 mm. The overall sensitivity of TAB in this cohort was 47% with specificity of 100%.

Sensitivity and specificity of TAB in patients with negative or inconclusive US

In the initial phase when the FTP was being established, there was some overlap in investigations, where patients had TAB despite a positive US or patients did not have an US requested and directly referred for TAB. The US was considered inconclusive when it showed evidence of abnormality in temporal artery such as thickening of the arterial wall, but no conclusive evidence of a halo sign.

Of the 104 patients, 15 were excluded, as 8 did not have an US prior to TAB and 7 had a positive US.

In the remaining 89 patients who had negative or inconclusive US, 44 had a final diagnosis of GCA of which 17 had a positive biopsy. The sensitivity of TAB as part of a sequential GCA FTP in patients with negative or inconclusive US was 39% with specificity of 100%.

Within this cohort, ten patients had an inconclusive US, of whom six had a final diagnosis of GCA. Four of the six (67%) had a positive TAB, thereby demonstrating a higher sensitivity in patients with inconclusive US, compared with those with negative US (34%).

In 7 patients with positive US, 4 had a positive TAB for GCA. Three patients with false positive US had a negative TAB for GCA (normal histology), of which one patient had diagnosis of bilateral cilio-retinal artery occlusion from carotid artery stenosis (TAB performed 3 days after steroid commencement) and one patient was diagnosed with medium-vessel vasculitis based on CT PET scan (TAB performed 14 days after steroid commencement). The third patient with history of head injury and subdural hematoma, had raised inflammatory markers and headache, hence investigated for GCA which was ruled out following a negative TAB (no steroids were started for this patient).

Impact of arterial specimen length on sensitivity and specificity of TAB

Table 2a shows the sensitivity and specificity of TAB depending on the post-fixation arterial specimen length in 5 mm increments (p = 0.468). The sensitivity was 14% in specimen length <15 mm, improving to 52% when specimen length was ≥15 mm (p = 0.155) (Table 2b).

Table 2 a: Comparing sensitivity and specificity of TAB based on post-fixation arterial specimen length, 5 mm increment (p = 0.47). b: Comparing sensitivity and specificity of TAB with post-fixation arterial specimen length <15 mm and ≥15 mm (p = 0.155).
Full size table

Impact of corticosteroid duration on sensitivity and specificity of TAB

The range of duration of corticosteroid use prior to TAB was 0–90 days with an average of 13 days.

The sensitivity of TAB when broken down by length of treatment with corticosteroid prior to biopsy, in weekly increments and 10-day increments is shown in Tables 3 and 4.

Table 3 Comparing sensitivity and specificity of TAB based on corticosteroid duration every 7 days (p = 0.159).
Full size table
Table 4 Comparing sensitivity and specificity of TAB based on corticosteroid duration every ten days (p = 0.07).
Full size table

Sensitivity of TAB was very similar in the first week (60%) and the first 10 days (59%) of initiating corticosteroids. No statistically significant difference was found in sensitivities if TAB was done within 7 days (60%) and 8–10 days (57%) (p = 0.33). The comparison of sensitivities was close to significance in the 10-day increment analysis (p = 0.07).

Analysis of clinical predictors for a positive TAB

We evaluated the patients’ clinical and laboratory characteristics in Table 1 to assess whether any of these increase the odds of having a positive biopsy.

Analysis was performed using a series of univariate logistic regression model, each one estimating how the odds for a positive biopsy are affected by the presence of a finding compared with its absence, using age and GCA diagnosis as additional covariates (Table 5). Predictors of a positive biopsy within this sequential GCA FTP included jaw claudication (OR = 5.40; p = 0.0057), elevated erythrocyte sedimentation rate (OR = 5.50; p = 0.013) and elevated C-reactive protein (OR = 23.7; p = 0.0043). Only three patients with positive biopsy had a normal C-reactive protein level.

Table 5 Univariate logistic regression model estimating how the odds ratio for a positive biopsy is affected by the presence of a clinical or laboratory characteristic compared to its absence.
Full size table

Discussion

This study demonstrates that despite the increasing popularity of US in GCA FTP, temporal artery biopsy continues to play an important role in diagnosing GCA when US does not provide definitive evidence. TAB becomes particularly important for patients who are not positive on US, partly for confirmation of diagnosis and appropriate management strategy, but also for eligibility for biologics use as per specialised commissioning in England. Long term treatment with corticosteroids is not without risk and since the symptoms for GCA can be non-specific, appropriate confirmation or refutation of diagnosis is an important aim in the management of these patients. Studies have also demonstrated that temporal artery US, showed more accuracy when performed on patients who had no more than one dose of steroid [1]. In a parallel study of the Coventry sequential GCA FTP, temporal artery US demonstrated a sensitivity of 50% (specificity was >96%), which improved further to ~56% in corticosteroid-naïve patients, with the results being from different time points (2013–2017) compared with our present dataset. In this cohort, 23% of patients (143/620) were diagnosed and treated as GCA. The majority (66%) of patients received their US on the day of their referral. Overall positive predictive value (PPV) for US over the 4 years was 84.9% with negative predictive value (NPV) being 86.7% and further improvement to 91.2% in CS naïve patients [10].

In one study, US has been shown to have a higher sensitivity (54%) than TAB (39%) and found to be more cost effective as a non-invasive procedure. However, more than half of the TAB specimens in this study did not meet the BSR guidelines which may have contributed to a lower diagnostic yield [1]. Very few studies on TAB have included only those patients, where the arterial specimen length is 10 mm or more, in keeping with the BSR guidelines and this can impact on sensitivity of biopsy. Though the updated BSR guidelines for management of GCA [11] and 2018 European League Against Rheumatism (EULAR) recommendation [12] state that the arterial specimen length should be at least 10 mm, our results demonstrate increased diagnostic sensitivity when specimen length was ≥15 mm. One study demonstrated that a TAB length cut-off point of ≥15 mm increased the odds of a positive TAB by 2.25 compared with a TAB < 15 mm (p = 0.003) [13] and another more recent study identified an ideal arterial specimen length to be a pre-fixation length of 1.5–2.0 cm, with longer biopsies unlikely to provide significant additional diagnostic yield [14]. Results from our study also demonstrate no significant improvement in diagnostic sensitivity with longer specimen length. One of the limitations in our study was that the pre-fixation specimen length was not recorded in the initial phase, although this was later rectified. Since then the authors routinely record it and prefer to obtain a pre-fixation specimen length of approximately 20 mm.

The accepted clinical practice is to perform a TAB within 2 weeks of commencing corticosteroids [15]. There are conflicting reports in the literature with some reports stating an 86% sensitivity after 4 or more weeks on corticosteroid treatment to others reporting the sensitivity reducing to 10% after 1 week on corticosteroid treatment [15, 16]. Another study reported that biopsies are more likely to yield positive features in patients with GCA prior to corticosteroid treatment but can still demonstrate positive features up to 4 weeks after steroid commencement [17]. The BSR guidelines on managing GCA recommend that an early TAB is desirable in patients with suspected cranial GCA, preferably within 7 days of initiating high-dose corticosteroid therapy [2]. In clinical practice having a time frame of up to 10 days from initiating corticosteroids provides adequate time to arrange for TAB to be performed by a surgeon with adequate expertise within elective operating sessions. The results from our sequential GCA FTP study show that TAB yielded a higher sensitivity up to 10 days after corticosteroid commencement. Thereafter, the sensitivity begins to decrease and is down to 0% after 20 days of corticosteroid commencement (which was trending towards statistical significance, p = 0.07).

Looking at clinical predictors for a positive TAB in GCA, one study demonstrated jaw claudication, and abnormal artery pulse to be the most predictive clinical features of a positive TAB [18]. Another study showed a statistically significant difference with respect to patient’s age and C-reactive protein level (CRP) (p < 0.001 and p = 0.012, respectively) between biopsy-negative and biopsy-positive groups, but no statistically significant difference in terms of sex, serum haemoglobin, platelet count, and erythrocyte sedimentation rate (ESR) [19]. Our study showed that patients presenting with jaw claudication, elevated ESR and CRP were more likely to yield a positive temporal artery biopsy within a sequential GCA FTP.

Luqmani et al. demonstrated that TAB was more specific for diagnosis of GCA than US (100% vs 81%), but no data was available regarding the TAB findings or the clinical diagnosis in patients with false positive US within this study [1]. Fernandez et al reported the frequency and causes of 14 patients with false positive US with presence of halo sign in patients without a diagnosis of GCA. In their study, 4.6% (14/304) of patients showed presence of the halo sign without final diagnosis of GCA. The final diagnoses in their patients were based on clinical findings after long term follow up (over at least 1 year), which included polymyalgia rheumatica (n = 4, 29%); atherosclerosis (n = 3, 21%); and non-Hodgkin’s T cell lymphoma, osteomyelitis of the skull base, primary amyloidosis associated with multiple myeloma, granulomatosis with polyangiitis, neurosyphilis, urinary sepsis and narrow-angle glaucoma (n = 1 each, 7%). They reported that some of these cases were false positives probably due to exam errors. However, others were due to diseases that increase the arterial wall thickness probably caused by cell infiltration and related oedema (such as non-Hodgkin’s T cell Lymphoma and ANCA-associated vasculitis) or by hypoechoic material deposit (as atherosclerosis and primary amyloidosis/multiple myeloma, although typically atherosclerotic lesions are echogenic rather than echolucent). In their study, 9 out of the 14 patients had TAB. Histology showed deposit of amyloid material in 1 patient with multiple myeloma. The biopsy was negative for GCA in the remaining 8 patients. They concluded that a very low proportion of patients without GCA will have the halo sign, and that the finding should be correlated with the patient’s clinical findings [20]. The results from our study demonstrated normal histology with TAB in the three patients with false positive US, whose final clinical diagnoses included carotid artery stenosis, medium-vessel vasculitis, and head injury with subdural hematoma. Hence clinicians should consider TAB, when they suspect an alternative diagnosis despite a positive US. One of the limitations with US is that its reliability can vary depending on the training of the operator. Hence appropriate temporal artery US training in the diagnosis of GCA should be provided, to reduce false positives and false negatives.

With the future of GCA management increasingly leaning towards sequential fast track pathways, this study provides much needed information about time frame to perform TAB after initiating corticosteroids and arterial specimen length, to enable appropriate patient management. When setting up fast track pathways for GCA, where imaging is the first investigation, clinicians need to be aware that the need for TAB will decrease significantly. Our study demonstrated that TAB was required in less than 10% of the referrals to the sequential GCA fast track pathway. This, in turn, requires a high-quality biopsy service as part of the GCA FTP, which can be delivered by a surgical team with adequate expertise.

In conclusion, our study demonstrates that TAB continues to play an important role in the diagnosis of GCA as part of a sequential fast track pathway, when US is negative or inconclusive and shows increased diagnostic sensitivity when performed by an experienced team with arterial specimen length ≥15 mm and done within 10 days of commencing corticosteroids.

Summary

What was known before

  • Several centres are establishing fast track pathways for management of GCA. Some studies suggest the use of temporal artery ultrasound as first line of investigation for GCA, as it is easy to obtain, cheap and non-invasive and to only perform temporal artery biopsy if ultrasound is negative.

  • Temporal artery biopsy is considered a gold standard investigation for GCA, when performed in a timely manner by experienced staff, with the BSR guidelines recommending an arterial specimen size of no less than 1 cm and ideally done within 1 week.

  • There is paucity of literature regarding the usefulness of temporal artery biopsy within GCA fast track pathways, especially in those patients where ultrasound is negative.

What this study adds

  • This is the first prospective study, which evaluated the role of temporal artery biopsy in a GCA fast track pathway, and showed that it continues to play an important role, when ultrasound is negative or inconclusive.

  • The study showed that the biopsy should ideally be performed within 10 days of starting corticosteroids and the post-fixation arterial specimen length should be ≥15 mm to increase the diagnostic sensitivity of the biopsy.