The role of PET/CT in disease activity assessment in patients with large vessel vasculitis

Purpose of review The aim of this article was to review the recent contributions on the role of PET in assessing disease activity in patients with large-vessel vasculitis (giant cell arteritis and Takayasu arteritis). Recent findings 18FDG (fluorodeoxyglucose) vascular uptake in large-vessel vasculitis at PET shows moderate correlation with clinical indices, laboratory markers and signs of arterial involvement at morphological imaging. Limited data may suggest that 18FDG (fluorodeoxyglucose) vascular uptake could predict relapses and (in Takayasu arteritis) the development of new angiographic vascular lesions. PET appears to be in general sensitive to change after treatment. Summary While the role of PET in diagnosis large-vessel vasculitis is established, its role in evaluating disease activity is less clear-cut. PET may be used as an ancillary technique, but a comprehensive assessment, including clinical, laboratory and morphological imaging is still required to monitor patients with large-vessel vasculitis over time.


INTRODUCTION
Giant cell arteritis (GCA) and Takayasu arteritis (TAK) are primary systemic large-vessel vasculitides (LVV) [1,2]. GCA typically affects individuals aged 50 or older [1], while TAK mainly affects young females [2]. 18 F-Fluorodeoxyglucose (FDG) positron emission tomography (PET), nowadays often co-registered with computerized tomography (PET/CT), can identify areas characterized by elevated glucose / FDG metabolism including inflammatory vascular lesions [3]. The role of PET in diagnosing LVV is reasonably well established [3,4]. In contrast, the usefulness of PET in monitoring LVV activity is more debated [3]. Part of the reason is that there is as yet no 'gold standard' to define LVV activity to which vascular FDG uptake could be compared; in clinical practice, the treating physician's judgment, which relies on clinical, laboratory and imaging findings, still represents the most common proxy to assess disease activity [5].
Over the last months, some studies have attempted to better clarify the role of PET in evaluating disease activity in patients with LVV. Herein, we reviewed the relevant papers and discuss their main findings.

ROLE OF PET IN ASSESSING DISEASE ACTIVITY
Because the intensity of FDG vascular uptake is considered to reflect vessel inflammation [3], a number of studies have attempted to correlate FDG vascular uptake with other parameters of disease activity. In addition, some studies have also looked at changes in FDG vascular uptake before and after treatment to define its sensitivity to change (Fig. 1).

Giant cell arteritis
A real-life study on 22 patients with GCA demonstrated that the global amount of basal FDG vascular uptake significantly decreased after 8 months (range 2-8 months) of treatment with tocilizumab (TCZ) [6]. At follow-up all patients were in clinical remission. To quantitatively assess the extent of vascular inflammation, a Total Vascular Score (TVS) ranging from 0 to 33, was used. However, among the abovementioned 22 patients, only 4 underwent PET/CT scans both before and after TCZ treatment and were recruited for this analysis. The authors concluded that PET/CT is valuable in LVV follow-up because it correlates well with clinically determined disease activity. On the other hand, the small number of patients enrolled with followup considerably hampers the generalization of the study's findings.
A recently published observational study was designed to monitor response to treatment in GCA by using PET/CT in addition to clinical and laboratory findings [7]. Eighty-eight patients with active, newly diagnosed LV-GCA were enrolled in this study. They satisfied the GiACTA trial inclusion criteria [8,9]; baseline PET was performed in all cases within 2 weeks of diagnosis. Twenty-seven patients received prednisolone as first-line therapy; 42 patients were treated with glucocorticoids þ methotrexate (MTX) while glucocorticoids þ TCZ were administered in 19 patients. At baseline 82/88

KEY POINTS
The role of PET in securing diagnosis of large-vessel vasculitis is well established.
The role of PET in assessing disease activity in largevessel vasculitis is more controversial. A comprehensive approach, which integrates clinical, laboratory and imaging data, should be used to monitor patients over time.
More studies with prospectively collected data are required to further elucidate the full potential of PET in monitoring patients with large-vessel vasculitis.

Pre-Therapy
Post-Therapy  patients had evidence of active vasculitis at PET scan. PETVAS, which is a new PET-based index aimed at quantitatively showing the global inflammatory burden [10], was used to assess treatment response in terms of vascular inflammation. The overall mean PETVAS decreased from 18.9 at baseline to 8.0 at follow-up. A reduction in global vascular uptake was seen in follow-up scans of all patients, regardless of treatment regimen. However, PETVAS reduction was significantly higher in patients treated with MTX (À12.3 units) or TCZ (À11.7 units) than in those receiving glucocorticoids alone (À8.7). PETVAS changes paralleled symptoms' resolution and reduction in acute-phase reactants' levels. According to the authors, GCA treatment allowed not only to achieve clinical remission and improvement of laboratory markers, but also to inhibit vascular inflammation. In this regard, Prieto-Pena et al. came to a different conclusion by highlighting a discrepancy between clinically determined and PET/CT determined disease activity in GCA patients treated with TCZ [11 & ]. In their observational study they included 30 patients with PET evidence of extra-cranial large vessel involvement (with or without concomitant cranial features). Imaging evaluation was assessed by using both a total vascular score (TVS) which represents the sum of Meller's score [12] in five selected arterial segments and a target-to-background ratio (TBR) of FDG uptake at the thoracic aorta. At baseline all patients were judged to be clinically active despite adequate glucocorticoid therapy. Therefore, TCZ treatment was started. After a mean follow-up period of 10.8 AE 3.7 months, there was a significant reduction in vascular uptake. TBR at the thoracic aorta decreased from a mean of 1.70 AE 0.52 to 1.48 AE 0.25 (P ¼ 0.005), while TVS fell from a mean of 4.97 AE 2.62 to 3.13 AE 1.89 (P < 0.001). At the end of the follow-up, 83.3% of patients were judged as being clinically inactive. Remission by PET/CT criteria was defined as a complete normalization of vascular uptake at follow-up; only 30% and 10% of patients had a normalization of TBR and TVS, respectively. Therefore, although there was an overall reduction in vascular uptake, no complete correspondence was seen between clinical remission and PET normalization. It is unclear whether this persistent low-grade vascular FDG uptake represents smoldering inflammation or post-inflammatory vascular remodeling.

Takayasu arteritis
Ma et al. have proposed a novel model to determine disease activity using PET in a retrospective cohort of 91 Chinese patients [13 & ]. All patients met the classification criteria of the American College of Rheumatology for TAK [2]. Sixty-five (71.4%) patients were na€ ıve, mean age was 37.2 years, while the female to male ratio was 2.6 : 1. Twenty-six (28.6%) of these patients were previously treated. Sixty-four (70.3%) were judged to have active TAK according to the physician's global opinion, of whom 57 (89.2%) also met the National Institute of Health (NIH) criteria for active disease [14]. Patients were assessed by PET as well as by other imaging procedures and laboratory markers [ERS, Creactive protein and interleukin (IL)-2 receptor]. The value of using 18 F-FDG-PET/CT to identify active disease was assessed using ESR as a reference. Disease activity assessment models were constructed and concordance index (C-index), net reclassification index (NRI), and integrated discrimination index (IDI) were evaluated. The best model to determine disease activity was found by combining the ESR, sum of SUV mean and IL-2R. The C-index of this model showed a substantial improvement compared with ESR alone (0.96 vs. 0.78, P < 0.01). This improvement was confirmed by positive reclassification [NRI, 1.63, 95% confidence interval (CI): 1.30, 1.97, P < 0.01) and an increased sensitivity (IDI, 0.48, 95% CI: 0.34, 0.62, P < 0.01). This disease activity model outperformed the NIH score [14].
The model proposed by Ma et al. [13 & ] is interesting, although it needs replication in a different, prospective cohort of patients. In addition, other biomarkers in addition to the IL-2 receptor should also be investigated. The choice of Ma et al. [13 & ] to use the SUV mean instead of the SUVmax has the advantage of averaging the activity score for a given arterial segment, thus avoiding overestimating disease activity within an arterial vessel, particularly when tracer uptake is patchy [15]. It is possible that SUV mean determination may also be more sensitive to change in assessing vasculitis activity [15], but this remains to be verified.
PET image acquisition is usually carried out one hour after contrast medium injection, but there is some evidence that image acquisition after a more prolonged time may increase PET sensitivity to detect vasculitis activity [16]. In this regard, De Souza Santos et al. [17] evaluated 20 patients with TAK fulfilling the classification criteria of the American College of Rheumatology [2] to establish the usefulness of PET with late-acquisition images in detecting disease activity in fully treated patients, a population in which the sensitivity of PET is known to be reduced [3]. Thirteen patients (65%) had active disease according to the NIH criteria [14]. This study showed that the mean SUVmax did not significantly discriminate between clinically active and inactive patients. No correlation of FGD vascular uptake with treatment, and only a weak correlation with the C-reactive protein (at the abdominal aorta) was found. This study thus provided evidence that PET with late-acquisition images may show signs of vascular FDG uptake even in fully treated TAK patients. However, the small number of patients enrolled and the lack of a control group, as well as of images acquired at an earlier time point, are important limitations of this study.
In previous report from our group, FDG uptake was observed in the aorta, but not in the pulmonary artery branch with wall thickening of a patient with TAK [18]. This suggested that PET might not be able to visualize the pulmonary arteries. To address the question whether PET lends itself to assess vascular disease activity in the pulmonary arteries from patients with TAK, Gao et al. prospectively recruited 29 TAK patients with pulmonary artery involvement and assessed them clinically, by morphological imaging, and by PET [19]. They found that PET was as sensitive as other imaging modalities to demonstrate active vascular disease (71.4% vs. 92.9%, P ¼ 0.250), thus confirming a role for PET in evaluating the pulmonary arteries. Pulmonary artery FDG uptake in PA positively correlated with ESR and C-reactive protein levels. In addition, to evaluate more specifically the usefulness of PET in monitoring the disease course, 8 patients had follow-up PET scans. Three patients with PET-active pulmonary artery vasculitis at baseline imaging showed a significant decrease in vascular FDG uptake at follow-up scans, whereas four patients without PET-active pulmonary arteries at baseline remained unchanged, and one patient who incurred a clinical recurrence of vasculitis showed a significant vascular increase in FDG uptake. The findings of this study may indicate that PET is not only able to diagnose active pulmonary artery vasculitis, but also monitor the disease course at this level.
Back in 2018, Grayson et al. developed a qualitative summary score based on global arterial FDG uptake, the PET Vascular Activity Score (PETVAS) [10]. In a study on 54 Chinese patients with TAK, Kang et al. [20] endeavored to evaluate the performance of PETVAS compared to SUVmax, inflammatory biomarkers and ITAS-2010 [21]. All patients fulfilled the American College of Rheumatology criteria for TAK [2]. Inter-observer agreement for PETVAS scoring was excellent (intraclass correlation coefficient 0.8). The biomarkers considered (ESR, C-reactive protein, and pentraxin-3), SUVmax, PETVAS, and ITAS-2010 scores were all significantly more elevated in patients with active compared with those with inactive disease. The area under the ROC curve of PETVAS (and pentraxin-3) were higher than those of SUVmax, ESR, C-reactive protein and ITAS-2010. A high correlation coefficient between PET-VAS and pentraxin-3 was observed. Longitudinal trends in alteration of PETVAS and pentraxin-3 at follow-up showed a good correlation with clinical progression or remission. These findings demonstrated that PETVAS was superior to the SUVmax in assessing the inflammatory status, while the high correlation between pentraxin-3 and PETVAS suggested that the former might be a useful biomarker to assess disease activity in TAK.

Large vessel vasculitis (giant cell arteritis and Takayasu arteritis)
Alessi et al. [22 && ] evaluated and compared disease activity over time in 50 GCA and 76 TAK patients using multimodal assessment combining clinical, laboratory, and imaging-based testing. Patients underwent standardized assessment, including PET, at enrollment and follow-up visits. Assessment of clinical disease activity was performed blinded to imaging findings and recorded as Physician Global Assessment (PhGA) on a scale of 0À10. Each PET study was subjectively interpreted as active or inactive vasculitis by an expert, blinded to clinical status. Global arterial FDG uptake was quantified by the PETVAS. Patients were stratified by disease duration at enrollment (0-2 years; 2-5 years; >5 years). Clinical disease activity was present in 33% of patients in the second-fifth year of disease and in 24% of patients evaluated >5 years after diagnosis. Active vasculitis by PET was observed in 66% of patients in years 2-5 after diagnosis and in 50% of patients enrolled >5 years after diagnosis. PETVAS were consistently higher in GCA than TAK in the early and later phases of disease and significantly decreased over time in GCA but not in TAK. Therefore, patients with GCA had a greater global burden of vascular inflammation at diagnosis and throughout the disease course. Vascular PET activity decreased over the duration of disease in GCA, but a similar reduction was not observed in TAK. Differences related to the increased presence of atherosclerotic lesions in GCA and later enrollment in the study of TAK patients in the course of the disease could partially explain these differences. However, these differences in longitudinal PET findings in GCA and TAK likely reflect fundamental differences in vascular biology between these conditions.
An important Italian retrospective study evaluated the accuracy of PET/CT and of PETVAS in assessing disease activity and the ability of PET-VAS in predicting relapses in a large single-center cohort of patients with LVV (51 large vessel GCA, 49 TAK) [23 && ].
Consecutive patients diagnosed with LVV who underwent at least one PET/CT scan between 2007 and 2020 were enrolled. The nuclear medicine physician's interpretation of each PET/CT scan (active/inactive vasculitis) was compared with disease activity clinical judgment (active disease/ remission). For each PET/CT scan, the PETVAS score was calculated. Four-hundred seventy-six PET/CT scans were performed in the 100 patients over a mean follow-up period of 97.5 months. Physician-determined PET/CT grading was able to distinguish between clinically active and inactive LVV with a sensitivity of 60% and specificity of 80.1%. The area under the curve (AUC) was 0.70. PETVAS was associated with disease activity, with an age and sexadjusted odds ratio for active disease of 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. The AUC was 0.73. PETVAS was not associated with subsequent relapses, with an age and sex-adjusted hazard ratio of 1.04 (95% CI 0.97, 1.11). This study clearly demonstrated that the visual PET/CT grading scale and PETVAS had moderate accuracy to distinguish active LVV from remission and that PETVAS did not predict disease relapses. Probably, the determination of a composite index that integrates symptoms, laboratory markers and morphologic and functional imaging, and physician and patient opinion remains a critical step in the assessment of disease activity in LVV.

Giant cell arteritis
The potential value of PET in predicting the risk of future relapse in LVV patients is controversial. While some authors have found an association between FDG vascular uptake during clinical remission and subsequent relapses [10], others have failed in demonstrating this relationship [23 && ,24]. In an interesting study, the medical records of 19 GCA patients, for whom a baseline PET/CT was available, were retrospectively reviewed to better clarify the prognostic implications of FDG vascular uptake [25]. Over a median follow-up of 15 months [interquartile range (IQR) 4.5-26.5], four relapses were recorded. No significant difference was found between baseline median SUVmax of patients who experienced a relapse )] and that of nonrelapsers [5.74 (IQR 4.98-7.99), P ¼ 0.92]. Similarly, PET of relapsers had a median ratio of SUV max aorta/liver of 1.24 (IQR 1.00-1.39), whereas it was 1.14 (IQR 0.90-1.40) in nonrelapsers (P ¼ 0.69). The authors concluded that baseline FDG vascular uptake was not useful in stratifying patients at higher risk of developing subsequent relapse.
In an Australian cohort, 21 patients newly diagnosed with GCA were prospectively evaluated for 12 months [26]. They all met the 1990 American College of Rheumatology (ACR) classification criteria for GCA [27], had a PET/CT scan performed within 72 h of prednisolone introduction and received a final clinical diagnosis of GCA. A TVS was calculated by adding Meller's score [12] in selected artery segments. Patients with a baseline TVS > ¼ 10 were no more likely to experience relapse than those with a baseline TVS < 10. Relapse rate was similar among the two groups. This study is interesting because it also addresses the role of PET/CT in monitoring treatment response. For fifteen out of twenty-one patients a PET scan after 6 months of treatment with glucocorticoids was available. The median TVS decreased from 14 (4-24) at baseline to 5 (0-10) at follow-up (P < 0.01). According to these findings, PET could be useful in monitoring response to treatment, but not in predicting relapse risk. However, these conclusions are hampered by the small size of cohort and the short duration of follow-up.

Takayasu arteritis
A study specifically endeavored to assess the role of PET in predicting relapses in TAK patients. Kwon et al. [28] included 33 patients with inactive TAK according to the NIH criteria [14]. All patients fulfilled the American College of Rheumatology criteria for TAK [2]. Relapse was defined as recurrence of clinically active disease following an inactive period and requiring therapy modification. Vascular FDG uptake was expressed using TBR, calculated as arterial SUVmax/ mean SUV in venous blood pool. Mean observation period was 4.5 years. Eight (27.3%) patients incurred a relapse. The only two factors associated on multivariable analysis with an increased risk of relapse were a raised ESR [hazard ratio (HR) 7] and TBR (HR 11.5). These findings suggest that TBR, together with the ESR, independently predict the relapse risk and are thus useful to stratify clinically inactive patients for the risk of relapse.
Another retrospective study from a single center tried to predict not only relapses (primary outcome), but also other clinically relevant parameters (sustained remission, development of new angiographic lesions and changes in the therapeutic regimen) [29]. Thirty-two patients with TAK were assessed at baseline and subsequently at regular intervals over a median of 84 months. Sustained remission was defined as lack of new manifestations of disease activity, no new angiographic lesions, and no treatment with glucocorticoids for at least 6 months. Eleven (34.4%) patients at baseline were judged to have active disease according to the modified NIH criteria [30]. Patients were stratified according to SUV max values !1.3 in at least one artery, because previously SUV max values !1.3 had shown an association with active disease [31]. This study demonstrated that patients with baseline SUVmax values !1.3 had an increased risk of relapses [odds ratio (OR) 5.66] and need to change immunosuppressive agents (OR 7.93). However, at multivariate analysis, no associations were found with the development of disease relapses and with new angiographic lesions. These results may suggest that PET may have an ancillary role in predicting relapses, but that morphological imaging remains necessary to monitor TAK patients.

ROLE OF PET IN PREDICTING ANGIOGRAPHIC PROGRESSION IN LARGE-VESSEL VASCULITIDES (GIANT CELL ARTERITIS AND TAKAYASU ARTERITIS)
Two recent studies, one retrospective, the second prospective, have evaluated whether vascular FDG-PET activity is associated with angiographic progression of vascular damage (stenoses/dilatations/ aneurysms) in vessel wall in LVV patients.
In the first study, Besutti et al. [5] evaluated 100 consecutive patients with LVV (49 with TAK and 51 with LV-GCA) with baseline PET/CT and morphological imaging (CT/MR angiography [CTA and MRA, respectively]) performed within 3 months. A total of 1206 vascular segments were examined. All available PET/CT and CT/MR scans were reviewed to assess PET-CT uptake (4-point semi-quantitative score), wall thickening, stenoses and dilations for 15 vascular segments. The associations of baseline PET score and CT/MR wall thickening with synchronous and incident stenoses/dilations at CT/MR performed 6-30 months from baseline were evaluated in per-segment and per-patient analyses. The authors found that a higher PET score was strongly associated with the synchronous presence of wall thickening at morphological CTA/MRA imaging (P < 0.001), while the association between PET score and synchronous stenoses/dilations was weaker (P ¼ 0.01). The presence of wall thickening was strongly associated with the presence of synchronous stenoses or dilations (P < 0.001). At least one follow-up CTA/MRA performed between 6 and 30 months from baseline combining imaging assessment was available in 28 patients, for a total of 299 vascular segments with a median interval time between baseline assessment and follow-up CTA/ MRA of 14.7 months. Seven new stenoses/dilations occurred, exclusively in TAK patients. Baseline PET score was strongly associated with incident stenoses/dilations (P ¼ 0.001), while baseline wall thickening was not (P ¼ 0.708). AUC for incident stenoses/dilations was 0.80 for PET score and 0.52 for wall thickening. This study showed that even if PET score and wall thickening are strongly associated, only PET score was a good predictor of new vessel-wall damage during the follow-up. However, this study did not include contrast enhancement edema and diffusion-weighted imaging as imaging biomarkers of vessel wall inflammation, therefore further studies focusing on MRA only may improve the predictive value of morphological imaging.
In the second study, Quinn et al. [16] evaluated prospectively 1091 arterial territories from 70 patients with LVV (TAK ¼ 38; GCA ¼ 32). All patients underwent baseline MRA or CTA of the aorta and primary branches as well PET; a follow-up study of the same image modality was performed at least six months after baseline per a standardized imaging protocol. Seventeen arterial territories were evaluated. Over 1.6 years of median follow-up, new lesions (stenosis, occlusion, or aneurysm) developed only in 8 arterial territories, exclusively in 5 patients with TAK. Existing arterial lesions improved in 16 territories and worsened in 6 territories. Most arterial territories without PET activity at baseline [787 out of 793 (99%)] remained unchanged over time by angiography. Only 24 (8%) of the 298 territories with PET activity changed over time, but the majority had baseline PET activity (80%). Within a patient, an arterial territory with baseline PET activity had 20 times increased odds for angiographic change compared to a paired arterial territory without PET activity (P < 0.01). Interestingly, concomitant wall morphologic changes (increased wall thickening and edema) further increased risk for angiographic change. Confirming the results of the previous study, the development of angiographic change was uncommon during the follow-up and when it occurred change was frequently preceded by the presence of FDG-PET activity in the arterial territory at baseline. Lack of FDG-PET activity was strongly associated with stable angiographic disease.
These two studies reinforce the need to use PET and morphological imaging as complementary assessments to evaluate disease activity and vascular damage over time in patients with LVV.
However, the use of PET score to guide treatment in LVV is still open to question, and more prospective studies are needed. Furthermore, in these two studies new wall damaged occurred only in patients with TAK, thus the results on PET as predictor of incident vessel damage are limited to TAK patients.