Imaging-guided stratification in giant cell arteritis: prognostic and therapeutic implications

Ultrasound

Its availability, low cost, and non-invasive nature make ultrasound a potentially useful tool for monitoring disease activity. Once treatment is initiated, it has been reported that the temporal arteries normalize faster than the extracranial arteries.^18–20 In Table 1, the percentage of pathological findings at the patient level is shown according to the evolution time of different imaging tests. In the earliest ultrasound studies in GCA, the temporal artery halo sign persisted for an average of 16 days. ²¹ However, the probes used at that time were much less sensitive than those currently available. Subsequent studies have shown that the sensitivity of ultrasound to detect abnormalities decreases as early as the second day after treatment initiation. ¹⁹ More recently, a prospective study confirmed this rapid decline, reporting a 27% relative reduction in temporal artery ultrasound sensitivity after just 10 days of glucocorticoid therapy. ²² In general, vasculitic abnormalities in the temporal arteries tend to resolve around 8 weeks after the initiation of therapy. ²³ Involvement of the axillary arteries may persist longer, with improvement typically seen after several months of treatment ¹⁸ (Figure 1). We cannot know for certain, given the difficulty of obtaining histological studies, whether these changes reflect persistent inflammation or vascular wall remodeling. In a longitudinal study of 40 patients with large-vessel GCA, follow-up ultrasound of the proximal arm arteries revealed changes suggestive of vasculitis in 70% of patients after a mean follow-up period of 39 months. ²⁴ Similarly, persistent increases in intima media thickness (IMT) of axillary arteries have been reported for up to 4 years in patients with long-standing disease, even in the absence of clinical relapse. ²⁵ However, the IMT cut-off values have been validated for the diagnosis of GCA,^26–29 and their usefulness in monitoring disease activity should be investigated further. In addition, caution is needed when interpreting the findings, especially in the presence of atherosclerosis. Over time, chronic lesions of the axillary arteries may exhibit a hyperechogenic appearance with multilinear bands, which has been recognized as a potential marker of structural damage rather than active vasculitis ³⁰ (Figure 2). These findings underscore the importance of integrating imaging morphology and temporal evolution, rather than relying solely on wall thickness or other quantitative parameters, to distinguish between active and chronic vascular changes.

OPEN IN VIEWER

Table 1.

Persistence of any pathological abnormalities at the patient level in different imaging tests for the assessment of disease activity in GCA.

Exam	Location	Percentage of positive findings	Follow-up	Number of patients	Reference
Ultrasound	Temporal arteries	94	3 days	16	Seitz et al. 31
		68	3 days	37	Hansen et al. 22
		53	10 days	47	Hansen et al. 22
		50	16 days	22	Schmidt et al. 21
		91	4 weeks	11	Seitz et al. 31
		30	8 weeks	47	Nielsen et al. 32
		92	8 weeks	12	Seitz et al. 31
		85	12 weeks	13	Seitz et al. 31
		77	24 weeks	13	Seitz et al. 31
		30	24 weeks	47	Nielsen et al. 32
		54	52 weeks	13	Seitz et al. 31
	Axillary arteries	57	3 days	37	Hansen et al. 22
		60	10 days	48	Hansen et al. 22
		57	8 weeks	47	Nielsen et al. 32
		66	24 weeks	47	Nielsen et al. 32
18F-FDG/PET	Extracranial arteries	100	3 days	10	Nielsen et al. 33
		36	10 days	14	Nielsen et al. 33
MRI	Temporal arteries	60	24 weeks	11	Christ et al. 34
		0	52 weeks	11	Christ et al. 34
	Aorta	67	12 weeks	9	Reichenbach et al. 35
		80	24 weeks	11	Christ et al. 34
		83	52 weeks	12	Christ et al. 34

¹⁸F-FDG/PET, ¹⁸F-fluorodeoxyglucose positron emission tomography; GCA, giant cell arteritis; MRI, magnetic resonance imaging.

OPEN IN VIEWER

Figure 1.

Longitudinal ultrasound scan of the right superficial temporal artery (a and b) in a patient with GCA at diagnosis (a) and after 2 months of treatment (b). Longitudinal ultrasound scan of the left axillary artery (c and d) in a patient with GCA at diagnosis (c) and after 6 months of treatment (d). Although the halo sign persists, note the appreciable decrease in IMT measurement following treatment. Improvement in IMT in axillary arteries is generally slower than in temporal arteries.

OPEN IN VIEWER

Figure 2.

Longitudinal examination of the right axillary artery in a patient with long-standing giant cell arteritis. Note the hyperechoic appearance of the arterial wall and the presence of multiple lines.

The GUSTO clinical trial has provided important evidence on the arterial changes observed by ultrasound following the initiation of treatment. ³¹ In this trial, an ultra-short course of glucocorticoids was administered (500 mg IV methylprednisolone for 3 days) to 18 patients, followed by IV tocilizumab on day 3 and weekly subcutaneous injections thereafter. A sharp decline in IMT was observed in the temporal arteries by days 2–3, followed by a rebound to baseline levels at week 4, and then a gradual decrease until week 52. In contrast, axillary arteries showed an initial IMT reduction, followed by a transient worsening and then a period of stability until week 24, after which a more progressive decline was detected. These findings may suggest that the axillary arteries have greater remodeling changes after inflammation than the temporal arteries, or that immunosuppressants are less effective at controlling inflammation in the extracranial arteries. Overall, the significance of subclinical ultrasound findings in patients considered to be in clinical remission remains to be determined. Specifically, it must be understood whether positive findings correspond to subclinical inflammation (and therefore deserve treatment escalation), vascular remodeling (which follows inflammation and precedes vascular damage and is probably less influenced by anti-inflammatory agents), or something else. ³⁶ The inclusion of other outcome measures such as hospitalization, risk of disability, or mortality, would have a major clinical impact on the evaluation of long-term subclinical findings.

Few published studies have compared ultrasound findings with markers of disease activity in GCA. In a retrospective study, De Miguel et al. ²³ showed that resolution of the temporal artery halo sign in 30 patients with new or relapsing GCA was associated with reductions in CRP and ESR during follow-up. More recently, Ponte et al. ¹⁸ conducted a prospective study involving 49 patients and demonstrated a correlation between temporal artery IMT and disease activity markers, including CRP, ESR, and the Birmingham Vasculitis Activity Score. Interestingly, no significant correlation was found for axillary artery IMT. In a subsequent prospective study involving 47 patients with GCA, Nielsen et al. ³² confirmed that temporal artery ultrasound-based scores correlated moderately with CRP and global disease activity measures, while large-vessel scores showed only weak or no correlation. Nevertheless, a validated definition of remission and disease activity in GCA is needed to assess the correlation between imaging findings and disease activity in future studies.

In recent years, composite ultrasound scores have been proposed to quantify the burden of arterial wall abnormalities, with some designed primarily for diagnosis and others for longitudinal monitoring.^37–41 The Southend Halo Score—which includes six temporal artery segments and both axillary arteries, assigning weighted scores based on IMT—was developed and shown to perform well as a diagnostic tool.^42,43 Although chiefly diagnostic, subsequent studies have assessed its utility for monitoring treatment response and quantifying changes in vascular inflammation over time.^32,44 More recently, the OMERACT ultrasound group has published the OMERACT GCA Ultrasonography Score (OGUS) to standardize the quantification of vascular abnormalities by ultrasound specifically for monitoring purposes. ⁴¹ OGUS is based on the same measurements in the same arterial segments as the Southend Halo Score, differing only in the method of calculation. OGUS has demonstrated both intra- and inter-observer reliability, as well as sensitivity to change. Nielsen et al., ³² in a study involving 47 patients with GCA that compared several quantitative ultrasound scores, demonstrated that OGUS was the index with the highest sensitivity to change. In a prospective 1-year cohort of 50 patients, Schäfer et al. ⁴⁵ observed improvements in IMT, number of affected arteries, and OGUS after treatment; relapses occurred in four patients, and at relapse mean IMT and OGUS were higher than at the preceding assessment.

Future options to improve disease monitoring, especially in extracranial arteries, include the use of contrast-enhanced ultrasound. In a proof-of-concept study involving 24 patients, Bergner et al. ⁴⁶ demonstrated that this technique had high sensitivity (91.7%) and specificity (100%) for detecting active extracranial disease when a >25% increase in the calculated contrast-enhanced area was observed. However, the inter-observer reliability of contrast-enhanced ultrasound of the axillary and carotid arteries for monitoring disease activity is low. ⁴⁷ Larger studies are needed to validate these results.

¹⁸F-FDG PET, MRI, and CT

Apart from ultrasound, ¹⁸F-FDG PET is the second imaging modality with the most evidence supporting its usefulness in monitoring GCA, especially for assessing extracranial involvement. However, its cost, exposure to ionizing radiation, and limited availability in many centers restrict its routine use for follow-up. Blockmans et al. ⁴⁸ studied 35 patients with ¹⁸F-FDG PET/CT at diagnosis, 3 months, and 6 months. They demonstrated that the Total Vascular Score (TVS) significantly decreased from diagnosis to 3 months, with no further significant reduction observed at 6 months. ⁴⁸ Grayson et al. studied 56 patients with large vessel vasculitis (LVV), including 30 with GCA, using serial ¹⁸F-FDG PET/CT at 6-month intervals. ¹⁸F-FDG PET/CT scans were interpreted as showing active vasculitis in most patients who were in clinical remission (41 out of 71 (58%)). Consequently, the specificity of ¹⁸F-FDG PET/CT to distinguish between imaging evidence of LVV and clinical remission was only 42%. ⁴⁹ In this study, the PET Vasculitis Activity Score (PETVAS) was developed to assess disease activity by summing the qualitative assessment of arterial ¹⁸F-FDG uptake compared to the liver across nine vascular territories (aorta, subclavian, carotid, and innominate arteries). Clinically active LVV was associated with a higher mean PETVAS (21.5 vs 12.2, p < 0.001), while lower scores were observed in patients in remission. The ability of PETVAS to discriminate between remission and relapse subsequently yielded a sensitivity of 68% and a specificity of 71% using a cut-off value of >20. On the other hand, the presence of atherosclerosis could also represent a confounding factor in patients considered to be in clinical remission, since age was associated with higher uptake. Other PETVAS cut-off points have been proposed to assess disease activity, showing similar performance. In a retrospective study of 100 patients with LVV, Galli et al. demonstrated an association of PETVAS with disease activity (OR for active disease 1.15, 95% CI: 1.11–1.19). A PETVAS ⩾10 provided 60.8% sensitivity and 80.6% specificity in differentiating between clinically active and inactive LVV. ⁵⁰ Eshagh et al., ⁵¹ in a small retrospective study of 38 patients with GCA, found no relapses in patients who had a PETVAS ⩽3 during follow-up (sensitivity 1, 95% CI: (1–1)), while relapse occurred in all but 2 patients who had a PETVAS ⩾7 (specificity 0.78, 95% CI: (0.44–1)).

The RIGA study aimed to assess the change in vascular inflammation in 88 patients with large vessel (LV)-GCA under different treatments using ¹⁸F-FDG PET/CT. PETVAS decreased from 18.9 to 8.0 units at follow-up in the overall population (p < 0.001). A numerically greater reduction in PETVAS was observed with combination treatment (methotrexate or tocilizumab vs glucocorticoid monotherapy). However, only 45.12% of follow-up scans were actually considered inactive, defined as the absence of vascular ¹⁸F-FDG uptake ⩾2 in any vascular territory. ⁵² These findings are in line with another study by Quinn et al., ⁵³ which showed a significant reduction in PETVAS after treatment, but also found a similar incidence of persistent active vasculitis (44%). These studies suggest that, while ¹⁸F-FDG PET is sensitive to changes in vascular inflammation, persistent uptake is common even in clinically quiescent disease, limiting its utility in routine monitoring. Emerging molecular imaging modalities may improve visualization of disease activity in GCA. Fibroblast activation protein inhibitor PET/CT can detect inflammation of large arteries, although persistent uptake has been observed even in remission, likely reflecting vascular wall remodeling and fibroblast activity. ⁵⁴ More recently, a novel inflammation-specific tracer, [⁶⁸Ga]Ga-DOTA-Siglec-9 PET/CT, which is expressed only during active inflammation, has demonstrated the potential ability to accurately depict GCA^55,56 (Figure 3). However, a recent pilot study including eight patients with GCA did not show higher arterial uptake compared to ¹⁸F-FDG uptake. ⁵⁷

OPEN IN VIEWER

Figure 3.

Representative [^{6 8} Ga]Ga-DOTA-Siglec-9 PET/CT images showing selected body regions in patients with GCA. (a) Treatment-naïve patient with relapsing GCA demonstrating intense vascular tracer uptake and periarticular signals consistent with concomitant polymyalgia rheumatica. (b) Different patient imaged after 1 week of prednisolone therapy, showing low vascular and periarticular uptake indicative of early treatment response.

MRI, although less studied than ¹⁸F-FDG uptake, has emerged as a promising tool for monitoring disease activity in GCA due to its ability to detect both cranial and extracranial arterial inflammation without ionizing radiation. In cranial GCA, vessel wall enhancement on MRI has been shown to decrease over time with treatment. Christ et al. ³⁴ reported complete resolution of cranial vessel wall enhancement at week 52 in 11 patients with GCA who underwent follow-up MRI, following a short-course glucocorticoid regimen combined with tocilizumab. Similarly, Rhee et al. ⁵⁸ observed a progressive reduction in vessel wall enhancement on serial orbital and cranial MRI scans performed at 1, 6, and 12 months in 11 patients with confirmed GCA. In extracranial arteries, however, abnormalities on MRI may persist despite clinical remission. ⁵⁸ Reichenbach et al. ³⁵ investigated the role of MRI in monitoring 30 patients randomized to receive TCZ or placebo, and found that late vessel wall enhancement persisted or increased in one-third of patients at week 52, despite complete clinical and laboratory remission. Taken together, these findings indicate that MRI signal may decline with therapy—especially cranially—yet persistence in LVs is common and complicates interpretation in routine follow-up. In a prospective comparison between MRA and ¹⁸F-FDG PET/CT, Quinn et al. ⁵⁹ showed that MRI better documented the anatomical extent of disease, but PET was more closely associated with clinical disease activity. Concordance between the two modalities was present in 68% of scans (Cohen’s kappa = 0.30), and clinical status was associated with disease activity by ¹⁸F-FDG PET/CT (p < 0.01) but not MRA (p = 0.70). Notably, 51% of patients in clinical remission showed active disease on both imaging modalities. To address variability in image interpretation, Froehlich et al. ⁶⁰ recently proposed the MRVAS scoring system to standardize assessment of wall thickening and enhancement, demonstrating excellent inter-reader agreement; however, its role in disease monitoring remains to be validated. Overall, MRI appears complementary to PET—offering superior structural mapping—while PET more consistently reflects metabolic activity. However, not all of these abnormalities detected by either modality correlate with clinical findings. The complementary nature of MRI and PET has prompted interest in hybrid ¹⁸F-FDG PET/MRI, with early pilot data from a study of 13 patients with LVV suggesting that combining structural and metabolic imaging may enhance disease activity assessment. ⁶¹

The use of CT is not recommended by EULAR for monitoring disease activity or detecting relapses, ⁵ but it could be useful for long-term monitoring of vascular damage, especially in regions with preceding vascular inflammation. A prospective study of 35 GCA patients undergoing CT during follow-up showed that contrast enhancement resolves in the majority of patients, but vessel wall thickening persists in 68%. ⁶² However, routine assessment of all GCA patients for vascular damage is not recommended.

GCA phenotypes detected by ultrasound

From a clinical perspective, at least four distinct phenotypes of patients with GCA can be identified, which are not mutually exclusive: (1) the cranial phenotype, with predominant involvement of the temporal arteries, may present with headache, scalp tenderness, or jaw claudication ⁶⁷ ; (2) the ischemic phenotype mainly includes patients with vision loss, usually due to anterior ischemic optic neuropathy, which can occur in 10%–25% of patients. ⁶⁸ In addition, although less frequently, GCA patients may present with cerebrovascular accidents ⁶⁹ ; (3) The extracranial phenotype, known as LV-GCA, predominantly affects the aorta and its major branches, and may cause nonspecific symptoms such as fever, constitutional syndrome, or fatigue, ⁷⁰ and (4) patients with GCA may present with symptoms consistent with polymyalgia rheumatica, even in the absence of GCA-related symptoms. ⁷¹ Imaging is part of the diagnostic process for GCA and helps us differentiate between the various phenotypes. In the case of LV-GCA, imaging confirmation of extracranial involvement is an essential requirement for diagnosis.

The LV-GCA phenotype has previously been associated with worse outcomes. However, most available data come from retrospective studies. Czihal et al., ⁷² in a retrospective study of 43 patients using baseline ultrasound, reported that patients with subclavian and axillary involvement experienced higher relapse rates (100% vs 35%) and greater need for glucocorticoid-sparing agents (56% vs 24%) compared to those with isolated cranial GCA. Sugihara et al., ⁷³ in another retrospective study of 139 patients, also found that the presence of LV-GCA detected by imaging at baseline was significantly associated with poor treatment outcomes at 24 and 54 weeks of follow-up (HR 3.54, 95% CI: 1.52–8.24). In contrast, two more recent prospective studies did not find baseline LV involvement to predict relapse. Monti et al., ⁷⁴ in a multicenter prospective cohort of 97 patients, and Haaversen et al., ⁷⁵ in a prospective study of 132 patients, both using vascular ultrasound, reported similar relapse rates across cranial, LV, and mixed GCA phenotypes. Another prospective study including 53 LV-GCA patients and 53 GCA patients without LV involvement found no differences in ischemic complications, glucocorticoid use, or treatment duration during follow-up. ²⁴ These findings suggest that while retrospective studies consistently associate LV-GCA with worse outcomes, prospective data remain limited and have yet to confirm a clear prognostic impact of LV involvement. Based on these results, recommendations advocate the same treatment strategies for both disease patterns. ¹ Nevertheless, imaging plays a central role in identifying LV-GCA, and its incorporation into patient phenotyping remains important for individualizing risk stratification as more robust longitudinal data emerge.

Quantification of arterial wall abnormalities by ultrasound

The quantification of arterial wall abnormalities by ultrasound at diagnosis may also provide relevant information for follow-up. In early studies based on the TABUL cohort, Monti et al. ⁴⁰ studied 207 patients and found no significant association between baseline ultrasound findings and 6-month outcomes related to disease activity/burden—including relapse, prolonged glucocorticoid use, and cumulative glucocorticoid exposure. In a separate prospective study, Aschwanden et al. showed in a prospective cohort of 42 patients that those with regression of vascular wall thickening during follow-up had lower baseline CRP and ESR compared to those with persistent thickening. This regression, however, did not translate into significant differences in relapse rates, cumulative glucocorticoid exposure, or need for DMARDs. ²⁰ In contrast, the first prospective study from the PROTEA cohort demonstrated in 49 GCA patients that a lower number of arterial segments with halo and reduced IMT in temporal arteries (but not in axillary arteries) were associated with a higher likelihood of being in clinical remission and a lower cumulative dose of prednisone. ¹⁸ Addressing relapse specifically, the subsequent prospective study by Schäfer et al. ⁴⁵ did not demonstrate a predictive value of OGUS for detecting relapses, likely influenced by the very low number of relapse events (4/50 patients).

In a recent retrospective study, Molina-Collada et al. included 76 patients with 6-month follow-up and showed that less early improvement in OGUS after initiation of treatment was associated with a higher risk of relapse at 6 months. This same research question was recently addressed by Monti et al. ⁷⁴ in a prospective study of 97 patients from the PROTEA cohort. They demonstrated that higher OGUS at diagnosis was associated with an increased risk of relapse within 12 months (incidence rate ratio (IRR) per 1 point increase in OGUS 1.85, 95% CI: 1.05–3.32). Additionally, OGUS score <1 within the first 3 weeks was negatively associated with subsequent relapses (IRR 0.44, 95% CI: 0.22–0.88), and not achieving this cut-off predicted shorter time to first relapse. In an ongoing international multicenter, prospective cohort, Molina-Collada et al. ⁷⁶ confirmed these results, showing that greater vascular inflammation at diagnosis and lack of normalization of vascular inflammation during the first 3 months of treatment confer higher relapse risk.

¹⁸F-FDG PET and MRI

In the study by Blockmans et al., ⁴⁸ which followed 35 patients with GCA over 6 months, baseline ¹⁸F-FDG PET/CT findings did not predict relapses during follow-up. Patients who relapsed showed similar early decreases in TVS compared to those who did not relapse. Sammel et al., ⁷⁷ using a similar approach with TVS in both cranial and LV territories at diagnosis, also found that PET activity did not distinguish relapsing versus non-relapsing disease. By contrast, Grayson et al., ⁴⁹ in a prospective study of 56 patients with LVV, found that those with high PETVAS (>20) during clinical remission were more likely to experience future clinical relapse than those with low scores (55% vs 11%; p = 0.03) over a median follow-up period of 15 months. Two additional studies, by Schönau et al. ⁵² and Quinn et al., ⁵³ also evaluated the prognostic role of PETVAS but failed to demonstrate consistent associations between PET findings—either at baseline or during follow-up—and future relapse. Complementing these PET data, Adler et al. ⁷⁸ reported in a randomized trial that MRI-derived mural thickening and enhancement did not correlate with relapse status, suggesting the limited utility of conventional MRI signals for relapse prediction. Taken together, current data do not yet support PET or MRI as standalone tools for relapse risk stratification, although PET-based composite scores (e.g., PETVAS) may hold promise pending further validation.