Repeated Cranial and Large‐Vessel Positron Emission Tomography/Computed Tomography Scans and the Association With Structural Aortic Disease in Giant Cell Arteritis: A Five‐Year Observational Study

Abstract

Objective

Giant cell arteritis (GCA) is characterized by cranial ischemia at diagnosis and late aortic structural disease. Repeated combined cranial and large‐vessel fluoro‐2‐deoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) scans were performed to assess the distribution of vasculitis activity over time and the relationship with clinical outcomes.

Methods

Patients were eligible if they were enrolled in a 64‐patient inception suspected GCA cohort in 2016 to 2017 and had a positive temporal artery biopsy and/or PET/CT scan at diagnosis. At five years, patients underwent a PET/CT scan, magnetic resonance aortogram, and clinical assessment. Scans were reported for overall metabolic disease activity and a visual FDG avidity grade at each vascular territory.

Results

Sixteen patients met inclusion criteria, and 11 attended the five‐year visit. Median age was 75 years, 73% were women, and all were in remission. At five years, 4 (36%) patients had aortic dilatation (range 40–43 mm), and five (45%) had globally active scans. Cranial artery activity resolved in all patients between diagnosis and five years, but aortitis developed in four patients who previously had PET‐inactive aortas. Disease‐modifying rheumatic drug (DMARD) use at five years was associated with scan inactivity (P = 0.02). There was a trend toward a higher mean aortic diameter in those with aortitis at five years (40.2 mm vs 36.0 mm, P = 0.06) but not those with aortitis at diagnosis.

Conclusion

Vasculitis activity changed from cranial and large vessel to exclusively large vessel by five years. This may explain the preponderance of early cranial and late aortic complications in GCA. The potential role of long‐term DMARDs to mitigate smoldering vasculitis warrants further study.

INTRODUCTION

Giant cell arteritis (GCA) is a chronic medium to large‐vessel (LV) vasculitis that commonly presents with ischemic cranial symptoms at onset and structural aortic disease many years after diagnosis. It is well recognized that the disease remains subclinically active in a significant proportion of patients, but the prevalence, vascular distribution, and clinical implications are not well understood.

Over the past decade, technological advancements have improved the resolution of ¹⁸F‐fluoro‐2‐deoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) and facilitated disease activity assessment in both cranial and LVs. PET/CT has a unique role among imaging modalities in its ability to directly visualize disease activity by detecting glucose use by inflammatory cells in the vascular wall. Among patients with a new clinical diagnosis of GCA, 56% to 86% have been shown to have activity in the cranial vessels (temporal, maxillary, occipital, facial, and vertebral arteries) and 67% to 88% in LVs, including 38% to 75% in the aorta. ¹ , ² , ³ , ⁴ , ⁵

The chronicity of the disease, particularly in the aorta and LVs, is well recognized. Surgical studies have demonstrated that patients with GCA can have histologically active ⁶ aortas many years after their diagnosis. Furthermore, clinical relapse is seen in up to 74% of patients within five years of disease onset. ⁷ Studies of PET/CT in GCA have similarly demonstrated persistent disease activity, but assessments have predominantly been at nonstandardized time points, less than two years from diagnosis, and none have assessed both cranial and LV activity. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³

An understanding of chronic cranial and LV activity is important when considering late clinical outcomes. Although a third of patients develop aortic dilatation or aneurysm on long‐term follow‐up, ¹⁴ , ¹⁵ this contrasts with a paradoxically low rate of vision loss of less than 1.5% in patients with established disease after initiation of treatment. ¹⁶ A large recent study reported that PET/CT–detected LV activity at diagnosis predicts aortic dilatation and aneurysm on long‐term follow‐up, ¹² but it is unclear if this risk relates to the presence of aortitis at diagnosis or the cumulative burden of aortic activity over many years. Furthermore, it is unclear if long‐term disease‐modifying rheumatic drugs (DMARDs) are effective in dampening inflammatory activity and modifying aortic structural outcomes. Regarding cranial vessels, temporal arteries were shown in one study to remain histologically active in 44% patients at 12 months ¹⁷ and ultrasound‐detected structural changes can persist in more than 20%, ¹⁸ but the prevalence of cranial disease activity beyond 12 months is unknown.

Our group reported the first prospective diagnostic assessment of combined cranial and LV PET/CT for GCA in 2019. ¹⁹ Patients in this cohort were followed prospectively with serial clinical, serological, and imaging assessments. This report summarizes the five‐year outcomes from this cohort with a focus on (1) the prevalence and distribution of cranial and LV PET activity, (2) the impact of PET‐detected aortitis on aortic structural disease, and (3) factors associated with the long‐term persistence of PET vascular activity.

MATERIALS AND METHODS

Patients in this five‐year cohort were originally recruited in 2016 and 2017 as a part of the GCA and PET scan study. ¹⁹ Sixty‐four consecutive patients suspected of having GCA were recruited from 13 centers and underwent a PET/CT scan less than 72 hours after starting glucocorticoids and before temporal artery biopsy. Sixteen of these 64 patients received a six‐month clinical diagnosis of GCA and had biopsy and/or PET/CT evidence of active vasculitis. This subgroup was eligible for the five‐year study. Patients provided written informed consent, and the project was approved by the local health district human ethics committee (HREC/16/HAWKE/68).

At the five‐year visit, patients underwent a standardized clinical assessment, blood collection, magnetic resonance imaging (MRI) of the aorta, PET/CT scan, and transthoracic echocardiogram (TTE). The clinical assessment included a questionnaire regarding GCA symptoms, medications, relapses, and vascular events. Findings were integrated with previously obtained data from four protocolized clinical reviews at three, six, 12, and 24 months postdiagnosis and six‐month PET/CT scans (these had been performed for patients with a positive temporal artery biopsy or CT aortitis at baseline). ²⁰ Blood was sent for C‐reactive protein (CRP) and erythrocyte sedimentation rate (ESR) testing.

The MRI of the aorta was performed using an electrocardiogram (EKG)‐gated, noncontrast protocol from the aortic arch to the abdominal aortic bifurcation using a 3T Siemens Magnetom Vida scanner (Siemens Healthineers). T1‐weighted (StarVIBE), T2‐weighted (HASTE, TSE‐Dixon), and steady‐state free precession cine images were interrogated by a single blinded cardiothoracic radiologist to measure the maximum internal and external wall diameters in short axis image planes perpendicular to the vessel long axis at six locations: the aortic root, midascending aorta, arch (isthmus), descending thoracic aorta (level with the midascending aorta), diaphragmatic hiatus, and renal arteries. Dilatation was defined by the external diameter at each aortic level exceeding the following: root and ascending aorta (40 mm), arch (35 mm), descending aorta (30 mm), and abdominal aorta (25 mm). ¹⁵

The five‐year PET/CT scans were performed on a Philips Ingenuity TF128 PET/CT scanner (Phillips Healthcare). Patient preparation, image acquisition, and reporting protocols were adopted from the protocol used for baseline and six‐month scans performed on a Siemens Biograph mCT scanner described previously. ¹⁹ Images were acquired from the vertex to diaphragm 60 minutes after intravenous injection of 100 megabecquerel of fluorine‐18 FDG. Arms were positioned by the side to allow better visualization of the head and neck vessels. Scans were independently reported by two PET‐experienced nuclear medicine physicians (EW and IH) with subjective overall assessment of the scan as globally active or inactive along with the intensity of FDG uptake in 18 cranial (temporal, maxillary, occipital, and vertebral) and LV segments. Specific criteria for a globally active scan were not defined, but rather, assessment was based on the distribution and intensity of FDG cranial and LV uptake. Vascular wall FDG uptake was compared with the blood pool avidity in the superior vena cava and graded as follows: 0, no uptake (less than or equal to blood pool); 1, minimal or equivocally increased uptake; 2, moderate or clearly increased uptake; and 3, very marked uptake. ¹² Discordant results relating to each patient's global scan assessment and the maximum FDG grade in any artery segment were resolved by a consensus read. Active aortitis was defined by the presence of grade 2 or higher avidity in the aorta in conjunction with an overall active scan report.

The noncontrast, non–EKG‐gated CT component of the PET/CT scan was used to obtain maximum external aortic root, midascending and proximal descending aorta dimensions. These were measured on the baseline and five‐year scans by one blinded reporter (IH) to assess for aortic growth. Due to cardiac motion and the lack of EKG gating, the aortic root measurements were expected to be less accurate than for MRI.

A focused TTE was performed as a bedside test to assess the aortic root and ascending aorta dimensions and compared against MRI. A blinded cardiac sonographer obtained images in the parasternal long axis plane and recorded dimensions at the aortic annulus, sinus of Valsalva, sinotubular junction, and tubular ascending aorta according to established criteria. ²¹ Measurements were recorded by a blinded imaging specialist cardiologist (GM).

Statistical analysis was performed using IBM SPSS version 29 and focused on the relationship among PET activity, aortic dilatation, and factors associated with persistent activity at five years. Comparisons between groups were performed using Fisher exact, the two‐tailed t‐test, and the independent samples Mann‐Whitney U‐test.

RESULTS

The five‐year visit was attended by 11 of 16 (69%) eligible patients, all of whom had a globally active PET/CT scan at diagnosis. Of the five who did not attend, three had died (gastric cancer, distal aortic thromboembolism, and unknown), and two declined for personal reasons. The median time from diagnosis was 5.6 years (range 4.0–6.3).

The median age of the cohort was 75 (64–85) years, and 73% were women. Patients were in clinical and serological remission with a median CRP of 1 mg/L (1–8 mg/L) and ESR of 10 mm/h (2–26 mm/h). This compared with a median CRP of 72 mg/L and ESR of 72 mm/h at baseline. Three were taking prednisone at doses ≤2.5 mg/day, and six were taking DMARDs (three methotrexate, one leflunomide, one tocilizumab, and one azathioprine). Ten patients (91%) had been on a DMARD at least once during their disease course, seven (64%) had relapsed, and one had a vascular event (lacunar stroke considered unrelated to GCA). Five patients had a history of cranial ischemic symptoms at diagnosis, including vision disturbance (five) and jaw claudication (four).

Despite all patients being in clinical remission, five patients (45%) had a globally active PET/CT scan at the five‐year visit. Each of these five patients had diffuse LV FDG activity involving the ascending, arch, and descending aorta and bilateral subclavian and axillary arteries. The pattern of activity was considered inconsistent with atherosclerosis. Importantly, four of these five patients did not have aortic activity on the baseline scan. As such, they had newly developed aortitis over the five‐year period while in clinical remission. Table 1 presents clinical and PET/CT characteristics for each of the 11 patients, and Figure 1 presents the corresponding FDG‐PET/CT oblique reformatted image of the thoracic aorta in the sagittal plane. A summary of the cohort based on disease activity at five years is presented in Table 2.

Table 1.

Patient characteristics at baseline, two years, and five years^*

		Baseline visit (at diagnosis)				Two‐year visit	Five‐year visit
Patient	Gender	Vascular risk factors	CRP (mg/L)	ESR (mm/h)	PET/CT active aortitis	DMARD	Age, years	Prednisone dose, mg	DMARD	CRP (mg/L)	ESR (mm/h)	Maximum aortic diameter, mm	PET/CT active aortitis
A	F	0	36	41	Yes	‐	75	‐	‐	3	10	39	Yes
B	M	1	280	116	No	‐	82	2	‐	1	12	43	Yes
C	M	1	45	24	No	MTX	74	‐	‐	3	8	42	Yes
D	F	2	72	91	No	MTX	85	‐	‐	1	26	35	Yes
E	M	2	8	14	No	LFL	71	‐	‐	1	5	42	Yes
F	F	3	111	69	No	AZA	81	2.5	AZA	4	20	35	No
G	F	3	63	120	Yes	MTX, LFL	64	‐	LFL	1	11	35	No
H	F	2	24	48	Yes	TCZ, MTX	78	‐	TCZ	1	2	36	No
I	F	1	140	72	No	MTX, LFL	83	2	MTX	7	7	42	No
J	F	0	191	83	No	MTX	72	‐	MTX	1	7	34	No
K	F	1	121	92	Yes	MTX	73	‐	MTX	8	10	34	No

^*Vascular risk factors (1 point for each of the following): smoker or ex‐smoker, hypertension, hypercholesterolemia, family history of stroke or myocardial infarction under 50 years, and diabetes.

AZA, azathioprine; CRP, C‐reactive protein; CT, computed tomography; DMARD, disease‐modifying antirheumatic drug; ESR, erythrocyte sedimentation rate; F, female; LFL, leflunomide; M, male; MTX, methotrexate; PET, positron emission tomography; TCZ, tocilizumab.

Figure 1.

Five‐year FDG‐PET/CT images in the sagittal plane of the thoracic aorta. CT, computed tomography; FDG, fluoro‐2‐deoxyglucose; PET, positron emission tomography.

Table 2.

Patient characteristics grouped by five‐year PET/CT scan activity^*

Characteristics	Full cohort (11)	Active (5)	Inactive (6)	P value
Age, median (range), years	75 (64–85)	75 (71–85)	75 (65–83)	0.66
Time from diagnosis to five‐year scan, median (range), years	5.6 (4.0–6.3)	5.7 (4.9–6.3)	5.4 (4.9–5.7)	0.43
Women, n (%)	8 (73)	2 (40)	6 (100)	0.06
Vision symptoms at diagnosis, n (%)	5 (45)	2 (40)	3 (50)	1.0
Jaw claudication at diagnosis, n (%)	4 (36)	2 (40)	2 (33)	1.0
Taking prednisone at 5 years, n (%)	3 (27)	1 (20)	2 (33)	0.58
Taking DMARD at 5 years, n (%)	6 (54)	0 (0)	6 (100)	0.02
Patients experiencing at least one clinical relapse 0–5 years, n (%)	7 (64)	4 (80)	3 (50)	0.50
Patients experiencing at least one clinical relapse 2–5 years, n (%)	2 (18)	1 (20)	1 (17)	0.73
CRP, median (range), mg/L	1 (1–8)	1 (1–3)	2.5 (1–8)	0.49
ESR, median (range), mm/h	10 (2–26)	10 (5–26)	8.5 (2–20)	0.53
Vascular events 0‐5 years, n (%)	1 (9)	0 (0)	1 (17)	0.73
Maximum aorta dimension, mean (range), mm	37.9 (34–43)	40.2 (35–43)	36.0 (34–42)	0.06
Patients with thoracic aortic dilatation, n (%)	4 (36)	3 (60)	1 (17)	0.20

^*DMARDs were methotrexate (three), tocilizumab (one), leflunomide (one), and azathioprine.

CRP, C‐reactive protein; CT, computed tomography; DMARD, disease‐modifying antirheumatic drug; ESR, erythrocyte sedimentation rate; PET, positron emission tomography.

In contrast to the LV findings, none of the 11 patients had cranial vessel activity at five years. There was progressive and sustained reduction in cranial activity over the observation period, noting that 9 of 11 (82%) patients had cranial activity on the baseline scan and 2 of 9 (22%) patients had cranial activity on the six‐month scan. Figure 2 illustrates the number of patients with cranial and aortic FDG avidity at baseline and five years, along with a patient example.

Figure 2.

Change in pattern of disease activity on PET by vascular territory from diagnosis to five years. PET, positron emission tomography.

The only clinical factor associated with scan inactivity was the use of DMARD therapy at five years. All six patients taking DMARDs had inactive scans, whereas all those not on therapy were PET active (P = 0.02). Other factors such as gender, low‐dose prednisone use, and a history of relapse did not have a significant association with activity.

Structural thoracic aortic disease based on the MRI was present in 4 of 11 (36%) patients, with dilatation to a maximum of 42 mm in the ascending aorta and 43 mm in the aortic arch. Figure 3 presents the five‐year MRI‐assessed diameters by aortic location. It is noteworthy from an aortic screening perspective that all patients with dilatation had abnormality in the root and/or ascending segment.

Figure 3.

Five‐year aorta diameters at each aortic segment for the 11 study patients (each line represents an individual patient).

The presence of aortic dilatation on MRI at five years was not associated with baseline PET aortic activity. Indeed, none of the four patients with dilated aortas had aortitis on the baseline PET scan. Interestingly, in three of these four (75%) patients, the dilatation was associated with de novo subclinical aortitis on the five‐year PET scan.

Although there was no association between baseline PET aortic activity and subsequent dilation on MRI at five years, there was a numerically positive correlation between five‐year PET aortitis and aortic diameter. Patients with aortitis at five years had a mean maximum aortic diameter of 40.2 mm compared with 36.0 mm (P = 0.06) for the six patients with inactive scans.

Mean five‐year aortic diameter growth based on the noncontrast, non–cardiac‐gated CT component of the baseline and five‐year PET/CT scans was 1.36 mm in the midascending aorta and 1.18 mm in the proximal descending aorta. These small changes are within the expected error bounds for aortic diameter measurement using nongated, noncontrast CT, and thus, no assessment could be made of growth based on the presence of aortitis on the baseline PET scan.

The TTE‐measured aortic root and ascending aortic diameters correlated well with corresponding MRI diameters, with a Pearson correlation of 0.91. All patients with a dilated aorta on MRI had a diameter exceeding 38 mm on TTE.

DISCUSSION

This is the first study to prospectively assess long‐term combined cranial and LV FDG‐PET/CT vascular findings in patients with GCA. It sheds light on the prevalence and distribution of vascular activity with the novel finding of resolution of cranial vascular activity and de novo aortic and LV activity developing in a significant proportion of patients.

A high proportion of our patients (43%) were found to have active PET/CT scans at five years despite clinical and serological remission. Although biopsy and imaging studies have found similarly high rates of vascular activity of approximately 40% within the first 24 months, ¹³ , ¹⁷ this study extends this timeline and demonstrates the chronicity of GCA. Interestingly, vascular activity at five years was isolated to LVs and the aorta with no residual cranial vascular activity. The trend away from cranial activity occurred early after treatment initiation, with only 22% of patients having cranial activity on the six‐month scan. The findings relating to early cranial activity and chronic LV activity correlate well with the clinical experience of GCA whereby aortic structural disease progresses over time, whereas late cranial ischemic events are extremely rare. ¹⁴ , ¹⁵ , ¹⁶ This was also seen in our patient cohort whereby five patients had cranial ischemic symptoms at the time of diagnosis, but none experienced cranial ischemic symptoms beyond 24 months.

The treatment and monitoring implications of LV and aortic disease activity at five years remain unclear. An important finding from this study was that all six patients who were taking DMARDs at five years had inactive scans, whereas the five patients not taking DMARDs had active scans. Although this indicates a possible role for long‐term DMARD therapy to minimize smoldering disease activity, due to our small cohort, these findings require confirmation in larger studies. Importantly, similar findings were reported in the resolution of vascular inflammation in patients with GCA study, ¹⁵ in which patients taking tocilizumab or methotrexate had numerically lower rates of active scans and lower PET vascular activity scores (PETVAS). It should also be noted that it is unclear whether reducing smoldering activity provides clinical benefit in terms of mitigating the development of structural aortic disease.

In contrast to a large recent study, ¹² we did not find an association between PET‐detected aortitis at baseline and dilatation at five years; indeed, none of the four patients with aortic dilatation had aortitis on the baseline PET. This would argue against the routine use of PET activity at diagnosis as the sole factor to guide structural aortic screening. Given that three of these four patients had de novo aortic activity on the five‐year scan, it could be the cumulative burden of aortitis that associates with dilatation rather than activity at a single time point. Because we did not have cardiac‐gated aortic measurements at baseline, we were unable to definitively assess the relationship between FDG activity at baseline and five years with aortic growth.

An additional point for discussion is the pathophysiology of FDG vascular uptake in patients in clinical remission. Although it has been theorized that it could reflect vascular remodeling or post‐vasculitis‐associated atherosclerosis, ¹¹ studies describing histopathological evidence of active aortitis in patients with GCA in remission would support the concept of smoldering vasculitis in these patients. ⁶

The key limitation of our study is the small cohort, limiting the generalizability of our findings and increasing the risk of type 1 and 2 errors. Although the small cohort may have resulted in missing important signals and overstating the impact of DMARDs in mitigating smoldering vasculitis, several findings were definitive, including the absence of association between baseline aortitis and dilatation at five years. Due to the lack of a cardiac‐gated CT or MRI at diagnosis to accurately calculate aortic growth, we cannot exclude the possibility that the patients who had aortic dilatation at five years had stable dimensions through the study period.

An additional limitation of the study that is generalizable for PET/CT in assessing arteritis is the subjectivity of reporting and lack of quantitative reporting criteria for disease activity. Semiquantitative grading systems comparing with liver are considered the current gold standard for LV vasculitis, but this is not feasible when assessing cranial vessels due to relatively low FDG activity in smaller vessels. Because we did not completely image the liver, we used a comparison to blood pool, which is considered an alternative background and has been used by other groups. ²² , ²³ We did not report or use standard uptake value thresholds because these are not considered to be sufficiently reproducible across studies and patients to confirm or refute disease activity. ²²

A key strength of our study was the well‐defined prospective cohort with all patients having PET/CT active GCA at baseline and progress scans at a predetermined timeframe. Other studies have included a mix of PET/CT, temporal artery biopsy, and clinically diagnosed cases. ¹²

The final important aspect of this study was the investigation of TTE as a low‐cost, bedside modality to assess for GCA aortopathy. Although GCA aneurysms may be seen in the ascending or descending aortic segments, it is rare to have isolated descending disease. ¹⁴ TTE is a well‐established modality to measure the aortic root and proximal ascending aorta, ²⁴ and we found that aortic dimensions correlated well with MRI. All patients with aortic dilatation had a TTE diameter >38 mm, and this may be a useful cut point for deciding which patients require cross‐sectional imaging.

CONCLUSION

In summary, this study provides important new insights into the longitudinal pattern of cranial and LV activity in GCA. A significant proportion of patients were found to have PET activity at five years, and the disease burden shifted from a mix of cranial and LV vasculitis at diagnosis to exclusively LV at five years. PET aortic avidity at baseline was not associated with dilatation at five years, and four patients developed de novo PET active aortitis over the follow‐up period. As such, the findings suggest that GCA is a dynamic disease, and we would caution against using a single PET assessment at diagnosis as the sole means to determine which patients to screen for late aortic structural disease. Our additional findings that the long‐term use of DMARDs may protect against smoldering vasculitis and that TTE could serve as a screening tool for GCA structural aortopathy warrant study in larger cohorts.

AUTHOR CONTRIBUTIONS

All authors contributed to at least one of the following manuscript preparation roles: conceptualization AND/OR methodology, software, investigation, formal analysis, data curation, visualization, and validation AND drafting or reviewing/editing the final draft. As corresponding author, Dr Sammel confirms that all authors have provided the final approval of the version to be published, and takes responsibility for the affirmations regarding article submission (eg, not under consideration by another journal), the integrity of the data presented, and the statements regarding compliance with institutional review board/Declaration of Helsinki requirements.

Supporting information

Disclosure Form: