Patterns of large-vessel lesions are associated with treatment outcomes in large-vessel giant cell arteritis

Objective To evaluate whether the distribution of large-vessel lesions (LVLs) in giant cell arteritis (GCA) is associated with poor treatment outcomes. Methods In a retrospective, multi-centric, nationwide registry of GCA patients treated with glucocorticoids between 2007 and 2014, 68 newly-diagnosed patients with LVLs were identied by imaging. Non-achievement of clinical remission by week 24 and/or relapse within 104 weeks were primarily evaluated. Factors inuencing the poor treatment outcome were analyzed using Cox proportional hazard modeling. Cumulative rates and median time to the rst event were analyzed by the Kaplan-Meier method and log-rank testing. -One-way analysis of variance used to compare three groups. Univariable and multivariable analyses of factors associated with an event of poor treatment outcome within 104 weeks were conducted using the Cox proportional hazard model. Age, any cranial symptoms, and polymyalgia rheumatica were selected as covariates of interest. GCs: glucocorticoids; LV: large-vessel; LVLs: large-vessel lesions.

group 3 included pulmonary, hepatic or mesenteric lesions in addition to carotid lesions. Baseline doses of GCs were not different across the groups. Mean time to achievement of low-dose GC treatment (prednisolone ≤5 mg/day) was also not signi cantly different between the groups. The cumulative rate of poor treatment outcome over the two years was 11.1%, 55.3% and 88.0% in the groups 1, 2 and 3, respectively; mean time to the events was signi cantly different among the groups. Multivariable analysis showed that the risk of poor treatment outcome was signi cantly higher in the group 3.

Conclusions
The distribution of LVLs was associated with treatment outcomes in large-vessel GCA. In addition to subclavian arteries, lesions in aorta and aortic branches other than subclavian arteries should be evaluated by imaging for large-vessel GCA. The pattern of LVLs determined by imaging should be considered when determining treatment strategies for GCA.

Background
Giant cell arteritis (GCA) is characterized by cranial symptoms, polymyalgia rheumatica (PMR), and largevessel lesions (LVLs) in the aorta or its branches. The subclavian artery is a key location for LVLs in GCA (1). Aortic dilatation or aneurysm is also an important clinical feature of GCA (2, 3) that is related to mortality (4,5). Aortic dissection or rupture of aneurysm was reported in previous studies (6-8), and aortic in ammation detected by imaging was associated with aortic dilatation (9). Based on previous reports on imaging of LVLs in patients with Takayasu arteritis (TAK) and GCA from the Diagnostic and Classi cation Criteria for Vasculitis cohort, it was shown that GCA patients were more likely to present with widespread lesions from the subclavian artery to the thoracic and abdominal aorta than TAK patients. In contrast, carotid lesions and/or lesions in renal, celiac or mesenteric arteries are infrequent in GCA patients (10).
Increasing dose of glucocorticoid (GC) for GCA led to GC-related complications (11,12), which were associated with increased mortality rate of GCA (13), and risk factors for prolonged treatment course of glucocorticoids (GCs) are important in determining the indication of combination therapy of tocilizumab and GCs (14)(15)(16). We reported that clinical characteristics of Japanese patients newly diagnosed with GCA are similar to those of Western countries (17). In addition, imaging revealed that LVLs were present in approximately half of all GCA patients, and that they were signi cantly associated with an increased probability of poor treatment outcome (17). We collectively hypothesize that the distribution of LVLs determines the response to treatment in patients with GCA. To test this hypothesis, we investigated whether the distribution of LVLs was associated with poor treatment outcome in patients with GCA using two-year follow-up data from our multi-center retrospective cohort study.

Database
The members of the Japan Research Committee of the Ministry of Health, Labour, and Welfare for Intractable Vasculitis (JPVAS) participated in this retrospective multi-center study, which was approved by the institutional review board of Tokyo Medical and Dental University (approval number: M2000-2084-01). The details were described in a previous study (17). The 68 newly-diagnosed GCA patients with LVLs were diagnosed between 2007 and 2014 in each facility based on imaging of LVLs with and without temporal arterial biopsy (TAB). Data for two-year follow-up post-initiation of GC treatment were collected, using pre-de ned case-report forms at 4, 8, 24, 52, 76 and 104 weeks after starting treatment.

Outcomes
The primary endpoint was an event related to a poor treatment outcome (classi ed as failure to achieve clinical remission by week 24 or relapse within 104 weeks). Clinical remission was de ned as absence of active disease as described in a previous report (17). Active disease at baseline was de ned as having clinical signs and symptoms in cranial and large-vessel areas, or suffering from PMR, and/or elevation of CRP levels. We distinguished persistent ischemic signs and symptoms (i.e., visual manifestations or claudication) from GCA damage at week 24 after starting treatment as follows: if the signs and symptoms of active disease at baseline persisted without worsening for six months or longer and the CRP level was normalized, these were considered to re ect the damage at week 24 after starting treatment. A patient was determined to have a relapse as described previously (17); if they had either one of the signs or symptoms newly appearing or worsening after achievement of clinical remission, or if they showed deterioration in the LVLs by imaging. Re-elevation of the CRP level above the limit of the normal range (<0.3 mg/dl) but without any clinical manifestations was considered as a relapse only if other causes of increased CRP such as infection were excluded and at that time treatment was intensi ed. Three investigators (HAU, HY, YN) independently con rmed clinical ndings at a time of relapse as described in the previous study (17).

Imaging of LVLs at baseline
The site investigators reported wall thickening, stenosis, aneurysm, or dissection in anatomical sites of lesions of aortic branches and aortic lesions based on the results of at least one of the following techniques: computed tomography (CT), CT angiography (CTA), magnetic resonance imaging (MRI), or magnetic resonance angiography (MRA). Fluorodeoxyglucose (FDG) uptake in the arterial wall on positron emission tomography (PET)-CT were also reported. Based on the imaging ndings, the GCA with LVLs were categorized into two phenotypes: GCA with aortic lesions (group 2) or without aortic lesions.
The latter were subdivided into large-vessel GCA (LV-GCA) with subclavian lesions (group 1: LV-GCA without aortic lesions) and that with lesions of aortic branches without subclavian lesions (group 3: atypical LV-GCA without aortic lesions).

Statistics
Student's t test and the Mann-Whitney test were used to compare continuous variables depending on their distribution, and the chi-squared test and Fisher's exact test were used for categorical variables. One-way analysis of variance was used to compare three groups. Cumulative rates and median time to the rst event of the poor treatment outcome or the achievement of low-dose GC (prednisolone ≤5 mg/day) treatment within 104 weeks were analyzed using the Kaplan-Meier method and the log-rank testing. Patients who did not achieve clinical remission by week 24 were deemed to have had an event of the poor treatment outcome at week 0 as described in the previous study (17,18). Univariable and multivariable analysis for associated factors with the event of poor treatment outcomes during 104 weeks were conducted by using the Cox proportional hazard modeling. All analyses were performed using SPSS version 26 (IBM, Armonk, NY, USA). All reported p-values are two-tailed and the level of signi cance is p <0.05.

Patients´ characteristics
Case report forms of 68 newly-diagnosed GCA patients with LVLs were assessed. All patients were of Asian ethnicity. The mean (standard deviation [S.D.]) age was 70.5 (7.0) years, and 70.6% were women.
Twenty-seven (39.7%) of the 68 patients were diagnosed by both positive TAB and positive imaging of LVLs, and 41 (60.3%) by positive imaging of LVLs only. Twenty-nine (42.6%) did not meet the ACR GCA classi cation criteria. Cranial lesions and PMR were observed in 41 (60.3%) and 23 (33.8%) of the 68 patients with LVLs, respectively. Wall thickening or FDG uptake in the arterial wall were common (70.6%), but the proportion of aortic aneurysms was low (8.8%) at baseline. Subclavian lesions were also frequently observed (57.4%), and 25% had stenosis of the aortic branches ( Table 1). The initial mean dose (S.D.) of prednisolone was 0.74 (0.26) mg/kg/day, and 20.6% received MTX for remission induction therapy, 4.4% cyclophosphamide, and 13.2% azathioprine. One-way analysis of variance was used to compare three groups.  One-way analysis of variance was used to compare three groups.   Figure 1A).
Methotrexate (MTX) was administered for initial treatment in 0% in group 1, 24.5% in group 2, 20% in group 3, and for relapse in 0% in group 1, 24.5% in group 2, and 10.0% in group 3 (Table 1) (Table 2). Cumulative rate of poor treatment outcomes in patients strati ed by the three groups of LVLs The cumulative rate of events of poor treatment outcome over the two-year follow-up was 11.1% for patients in group 1, 55.3% in group 2, and 88.0% in group 3. Mean time to an event of the poor treatment outcome was signi cantly different within each pair of the groups ( Figure 1B)

Discussion
In the current study, approximately 70% of GCA patients with LVLs con rmed by imaging exhibited aortic lesions. One half of these patients with aortic lesions (group 2) failed to achieve clinical remission by 24 weeks or relapsed after achieving clinical remission within the two-year follow-up. Patients with subclavian arteritis without aortic lesions (group 1) had the better treatment response than the group 2, and atypical large vessel-GCA without both aortic and subclavian artery involvement (group 3) had the worst prognostic phenotype of LV-GCA. Interestingly, the mean age of onset in the atypical LV-GCA (group 3) was older compared to the typical LV-GCA with subclavian lesions (group 1) or aortic lesions (group 2), and cranial lesions and PMR were common ndings in the atypical LV-GCA (group 3).
Prolonged aortic in ammation might contribute to slowly progressive vascular damage in GCA patients who have prior LVLs (9,19,20), and serious complications in the aortas of patients with GCA could result in increased mortality (4,6,7,20,21). LVLs were predictive factors of new cardiovascular events of GCA (22). Our JPVAS cohort showed that LVLs were associated with an increased probability of poor treatment outcome of GCA, especially when present in the aorta rather than its branches (17). Notably, the present study revealed that GCA patients with aortic lesions, mainly due to aortic wall in ammation, were more resistant to initial therapy than patients with subclavian arteritis without aortic lesions. Our results indicate that the aortic wall in ammation served as a factor predicting poor response to initial treatment. Subclavian / axillary artery was easier to assess by ultrasonography than aortic lesions, but the aortic wall in ammation by MRI or PET-CT should be also assessed to determine disease activity of GCA (23,24).
Patients with subclavian artery involvement with or without aortic lesions had a higher relapse rate than in patients with cranial GCA without LVLs (25). Our JPVAS cohort found a better treatment response in cranial GCA without LVLs (17) and subclavian arteritis without aortic lesions (group1). However, arm claudication due to subclavian artery stenosis at baseline was most frequently counted as damage at week 24 in the present study ( Figure 2). Recent imaging studies also showed arm claudication was associated with subclavian artery damage, but not in ammation (26). This is in line with the results of the evaluation of damage in previous prospective studies (27).
Recent meta-analysis showed the overall prevalence of GCA relapse was about 50 % with a high heterogeneity, and shorter GC regimens in randomized controlled trials was associated with the higher relapse rate (28). In our observational studies, patients with aortic lesions (group 2) had similar relapse rate with the meta-analysis, and group 1 and group 3 had clearly different relapse rate, even though mean time to achievement of low-dose GC treatment was similar among 3 groups.
Previous study has shown that new lesions by imaging may appear even in the absence of typical signs and symptoms, underscoring the necessity of comprehensive vascular imaging (29). Vessel wall signals by MRI (30)  Gribbons et al. sought to identify subsets of GCA, and showed that GCA with LVLs was divisible into two groups based on the involvement of temporal artery (37). TCZ improved clinical outcomes not only in GCA patients with cranial lesions but also in patients who presented with LVLs and PMR without cranial lesions (38). The present study showed proportion of cranial lesions were similar between LV-GCA patients with and without poor treatment outcome.
Group 3 in the present study was considered atypical LV-GCA because the patients did not have both aortic and subclavian artery lesions. Interestingly, the pattern of LVLs in group 3 had imaging features similar to TAK (10), while the mean age of onset was older compared to the other two groups, and high proportions of cranial lesions and PMR were present. A recent prospective study also documented progression of arterial stenosis or dilation in the relatively infrequent vascular lesions of GCA (29). Thus, the older age of onset and higher proportion of cranial lesions in group 3 were likely to be those of cranial GCA (1,25), but these had atypical LVLs.
There are some limitations to this study associated with its retrospective design, as described in the previous report (17). First, there was no uniform treatment protocol and imaging examination for assessment of LVLs in the 3 groups. Second, there was a potential indication bias in selecting treatment because the attending physician might intensify treatment for patients with atypical lesions, such as pulmonary vasculitis. However, we did not nd a signi cant difference among groups regarding the dose of GCs and the choice of immunosuppressive drugs. Third, although there might be some imprecision due to the relatively small sample size, signi cant differences among the three groups were obvious.

Conclusions
The distribution of LVLs was associated with treatment outcomes in large-vessel GCA. In addition to the conventionally screened subclavian artery, involvement of aorta and other large arteries should be evaluated in patients with GCA.

Declarations
Ethics approval and consent to participate This study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Epidemiological Research in Japan. The institutional review board of Tokyo Medical and Dental University as main center (approval number: M2000-2084-01) and the additional institutions approved our retrospective study without personally identi able information. Patients received posters informing the present study, or the posters were displayed in outpatient clinic of each facility.

Consent for publication
Not applicable Availability of data and materials All of the data supporting the conclusions of this article are included within the article.