Content area
Giant cell arteritis (GCA) is the most common form of vasculitis in western countries. ^sup 18^F-FDG PET has been shown to be a valuable tool for the diagnosis of extracranial GCA, but results of studies are inconsistent due to a lack of standardized ^sup 18^F-FDG PET criteria. In this study, we compared different semiquantitative approaches using a controlled design to define the most efficient method.
All patients with biopsy-proven GCA who had undergone an ^sup 18^F-FDG PET/CT scan in our PET unit were reviewed and matched with a control group based on age and sex. Different semiquantitative arterial (ascending and descending thoracic aorta and aortic arch) to background (liver, lung and venous blood pool) SUV ratios were blindly compared between GCA patients and matched controls.
We included 11 patients with biopsy-proven GCA cases and 11 matched controls. There were no differences between the groups with regard to body weight, injected radioactivity, blood glucose level or CRP. The arterial to venous blood pool ratios discriminated the two groups better than other methods when applied to the aortic arch and the descending thoracic aorta (p<0.015). In particular, the highest aortic to highest blood pool SUV^sub max^ ratio, when applied to the aortic arch, provided optimal diagnostic performance (sensitivity 81.8 %, specificity 91 %, AUC 0.87; p<0.0001) using a cut-off value of 1.53.
Among all tested ^sup 18^F-FDG PET/CT methods, the aortic to blood pool SUV^sub max^ ratio outperformed the liver and lung ratios. We suggest the use of this ratio for the assessment of aortic inflammation in GCA patients.[PUBLICATION ABSTRACT]
Eur J Nucl Med Mol Imaging (2014) 41:155166 DOI 10.1007/s00259-013-2545-1
ORIGINAL ARTICLE
Towards an optimal semiquantitative approach in giant cell arteritis: an 18F-FDG PET/CT case-control study
Florent L. Besson & Hubert de Boysson &
Jean-Jacques Parienti & Grard Bouvard &
Boris Bienvenu & Denis Agostini
Received: 19 February 2013 /Accepted: 13 August 2013 /Published online: 6 September 2013 # Springer-Verlag Berlin Heidelberg 2013
AbstractPurpose Giant cell arteritis (GCA) is the most common form of vasculitis in western countries. 18F-FDG PET has been shown to be a valuable tool for the diagnosis of extracranial GCA, but results of studies are inconsistent due to a lack of standardized 18F-FDG PET criteria. In this study, we compared different semiquantitative approaches using a controlled design to define the most efficient method.
Methods All patients with biopsy-proven GCAwho had undergone an 18F-FDG PET/CT scan in our PET unit were reviewed and matched with a control group based on age and sex. Different semiquantitative arterial (ascending and descending thoracic aorta and aortic arch) to background (liver, lung and venous blood pool) SUV ratios were blindly compared between GCA patients and matched controls.
Results We included 11 patients with biopsy-proven GCA cases and 11 matched controls. There were no differences between the groups with regard to body weight, injected radioactivity, blood glucose level or CRP. The arterial to venous blood pool ratios discriminated the two groups better than other methods when applied to the aortic arch and the descending
thoracic aorta (p <0.015). In particular, the highest aortic to highest blood pool SUVmax ratio, when applied to the aortic arch, provided optimal diagnostic performance (sensitivity81.8 %, specificity 91 %, AUC 0.87; p <0.0001) using a cut-off value of 1.53.
Conclusion Among all tested 18F-FDG PET/CT methods, the aortic to blood pool SUVmax ratio outperformed the liver and lung ratios. We suggest the use of this ratio for the assessment of aortic inflammation in GCA patients.
Keywords PET/CT .18F-FDG .Giantcellarteritis .Vasculitis
Introduction
Giant cell arteritis (GCA) is the most common form of vasculitis in western countries and primarily affects women over 50 years of age [1]. The pathophysiology of this disease has been described [2, 3] and is thought to result from the activation of adventitial dendritic cells in the temporal arteries. The wide range of clinical manifestations renders the diagnosis of GCA difficult [4]. Temporal artery biopsy (TAB), which is one of the five classification criteria of the American College of Rheumatology (ACR) [5], remains the gold standard for assessing diagnoses. The diagnosis is established if three of the five criteria are met. Nevertheless, TAB has a high false-negative rate, ranging from 10 % to 40 % [69]; furthermore, the involvement of the thoracic aorta or the main branches, which are involved in 1015 % of patients with GCA [10, 11] and more than 45 % of newly diagnosed patients with GCA [12, 13], is associated with a negative TAB in 50 % of reported cases [14]. Additionally, more than 10 % of GCA patients present with atypical systemic symptoms [15]. This context has promoted the development of noninvasive imaging modalities, especially for the diagnosis of extracranial GCA such as aortitis.
Electronic supplementary material The online version of this article (doi:http://dx.doi.org/10.1007/s00259-013-2545-1
Web End =10.1007/s00259-013-2545- 1) contains supplementary material, which is available to authorized users.
F. L. Besson (*) : G. Bouvard : D. AgostiniDepartment of Nuclear Medicine, CHU Caen, Avenue de la Cote de Nacre, 14000 Caen, Francee-mail: [email protected]
H. de Boysson : B. BienvenuDepartment of Internal Medicine, CHU Caen, Caen, France
J.<J. ParientiDepartment of Biostatistics, CHU Caen, Caen, France
D. AgostiniUniversity of Caen Lower-Normandy, EA 4650, Caen, France
156 Eur J Nucl Med Mol Imaging (2014) 41:155166
18F-FDG PET has been widely used in oncology. Tumour cells are well known to show increased glucose consumption [1618]. During the last decade, 18F-FDG PET has been evaluated as a diagnostic tool in GCA. Blockmans et al. observed a smooth linear and segmental 18F-FDG uptake pattern in the subclavian arteries and aortas of GCA patients [19]. Several authors have proposed various 18F-FDG PET diagnostic criteria, ranging from visual analysis [20] to semiquantitative evaluations based on SUVmax [21, 22] or SUVmax to background ratios [23, 24]. Despite a lack of standardized evaluation criteria, 18F-FDG PET global diagnoses appear to be valuable in extracranial GCA, as was illustrated by our previous systematic review and meta-analysis [25]. For a potential standardization of 18F-FDG PETcriteria in the assessment of extracranial GCA, the most relevant methods have yet to be compared.
We performed a retrospective case control study to identify the most relevant semiquantitative methods [25] for assessing aortic wall inflammation in extracranial GCA, which could be applied as a standard in the future.
Materials and methods
Study population
All patients with a diagnosis of temporal biopsy-proven GCA who underwent an 18F-FDG PET/CT scan in our PET unit between 2008 and 2012 for extracranial evaluation were reviewed. Only those who fulfilled the following criteria were included in the study: the ACR criteria were met and histo-logical evidence of vasculitis was found by FDG PET/CT, and serological C-reactive protein (CRP) results concomitant with the PET/CT acquisitions were available, as well as the clinical and therapeutic statuses. Each included GCA patient was then matched with a corresponding control patient based on age and sex (Fig. 1). Patients from the control group had no history of GCA or polymyalgia rheumatica and were referred to our PET unit for neoplastic or infectious diseases. Among all potential reviewed controls, only those with complete hybrid PET/CT data, serological CRP results concomitant with the PET/CT acquisitions and no liver or extensive lung diseases were included in the study (Fig. 1). The local ethics committee approved this retrospective study.
Image acquisition and reconstruction
All included patients underwent a hybrid 18F-FDG PET/CT scan on a Siemens Biograph 6 TrueV-HD. After a delay of 1 h following injection, a noncontrast transmission CT scan was performed for attenuation correction and coregistration. Next, a PET emission scan was obtained with 3 min per bed position in 3-D mode. Images were acquired according to standard procedural guidelines (46 h of fasting prior to injection of 4 MBq/kg
of FDG and glycaemia levels below 1.8 g/L). The 3-D PET images were reconstructed using an iterative algorithm (OSEM mode, four iterations and eight ordered subsets). Coregistration between the PET and CT acquisitions was automated.
Image analysis
Before retrospective image analysis, all enrolled patients (GCA and controls) were pooled and numbered for anonymization in to alphabetical order. The anonymized 18F-FDG PET/CT results were randomly re-evaluated with the same Siemens 3-D viewer software. For each patient, the thoracic aorta was divided into three main parts: the ascending aorta, aortic arch and descending thoracic aorta. An arterial region of interest (ROI) containing the arterial wall and the lumen was drawn around the aorta on each slice of the coregistered transaxial PET/CT images to obtain an arterial SUVmax for each
PET/CT axial slice. This procedure was performed for the three thoracic aortic territories (Fig. 2a). Next, we compared three types of semiquantitative approaches, two that have been previously described for GCA [23, 24] and one that has been described for atheroma but never applied to GCA [26] (Fig. 2b).
The first type of approach, adapted from that described by Hautzel et al. [24], included two methods (A and A) in which the arterial activity was normalized to the liver uptake (arterial to liver ratio): in method A, the highest SUVmax of the arterial
territory was normalized to the highest SUVmax of the whole
liver (the SUVmax of the liver was determined from fitted ROIs that were drawn on each fused axial slice image of the liver); in method A, the average SUVmax of the arterial
territory was normalized to the average SUVmax of the liver
(the average SUVmax is the mean value calculated from all SUVmax values of each slice image).
The second type of approach, adapted from that described by Moosig et al. [23], included two methods (B and B) in which the arterial activity was normalized to the lung uptake (arterial to lung ratio): in method B, the highest SUVmax of the arterial
territory was normalized to the SUVmax of the lung; in method
B, the average SUVmax of the arterial territory was normalized to the SUVmax of the lung. The lung 2-D ROI was obtained from a 2-cm-diameter axial 2-D ROI drawn in a low-activity region of the lung. The lung region was carefully selected to be far from any suspicious noisy activity (e.g. no hyperactive mediastinal lymph nodes or parenchymatous lesions close to the ROI).
In contrast to the previous two types, the third type of approach (methods C and C), adapted from that described by Rudd et al. [26] and based on the arterial to blood pool uptake ratio, has never been applied in GCA; however, the approach was tested in atherosclerosis patients to evaluate arterial wall inflammation and has been proven to be robust and highly reproducible [26]. In method C, the highest arterial SUVmax was normalized to the highest venous SUVmax. In method C, the average arterial SUVmax was normalized to the
Eur J Nucl Med Mol Imaging (2014) 41:155166 157
Fig. 1 Flow chart of the selection process
venous blood pool activity. This blood pool activity was obtained by averaging several venous fused axial ROIs (eight per patient in our study) that were drawn on the internal jugular veins if the targeted arterial territories were supradiaphragmatic; in our study, the right internal jugular vein was used.
Additionally, a CT evaluation of the arterial wall, adapted from that described by Tatsumi et al. [27], was performed in all patients to evaluate the degree of calcification in each territory. For each arterial territory, the greatest calcified protrusion was retained and was defined as an arterial attenuation of more than 130 Hounsfield units [28], measured into the lumen from the aortic wall. Four grades of classification were used according to the degree of thickening: 0 no calcification, 1 calcification thickening 4 mm, 2 calcification thickening 4 to 8 mm, 3 calcification thickening >8 mm [27].
Statistical analyses
Continuous variables are expressed as means SD. The two-tailed Wilcoxon matched-pairs signed ranks test was used to evaluate differences between continuous paired variables. For unpaired data comparisons, the two-tailed Mann-Whitney test was used. Categorical variables were compared using the 2 test. Correlations between quantitative values were performed using the two-tailed Spearmans rank order test. P values less than 0.05 were considered significant for all statistical analyses.
ROC analyses of each of the approaches were also performed. All statistical analyses were performed with GraphPad Prism software (version 6), and ROC curves were obtained using GraphPad and MedCalc software.
Results
Selection and characteristics of the population
Between 2008 and 2012, 255 patients underwent 301 18F-FDG PET/CTscans for inflammatory conditions in our PET unit, and33 GCA patients (61 18F-FDG PET/CT scans) were identified (Fig. 1). The GCA patients underwent FDG PET/CT because they were suspected of having aortitis or to evaluate the vascular wall inflammation in those already under treatment for known aortitis. For the retrospective analysis, only 11 patients with biopsy-proven GCA (one 18F-FDG PET/CT scan per patient) fulfilled the selection criteria. The selection process is summarized in Fig. 1. Eight (73 %) of the included GCA patients were receiving corticosteroids at the time of the 18F-FDG PET/CT scans; for 1 patient, corticosteroids were received in combination with methotrexate, and for another, cyclophosphamide. The 3 remaining patients were free of treatment for at least 6 months. From the PET unit database, we extracted 11 control patients with available concomitant CRP data, who were
158 Eur J Nucl Med Mol Imaging (2014) 41:155166
matched with the GCA patients on the basis of age and sex. Eight of the 11 controls underwent 18F-FDG PET/CT for malignancies, which included 1 cancer of unknown origin, 1 plasmocytoma, 1 vesicular cancer, 2 lung cancers, 1 grelic lymphoma and 2 breast neoplasms. The remaining 3 controls underwent 18F-FDG PET/CT for characterization of a monoclonal IgG disease, a focal lung infection and a suprarenal nodule. There were no differences in CRP levels (mg/L) or body weight (kg) in either group (Table 1).
Imaging parameters
Glycaemia, mean injected doses of 18F-FDG, CT dose indexes and doselength products were not significantly different between the two groups (Table 1). The delay times between
18F-FDG injection and image acquisition were slightly different(68.5510.96 min versus 60.364.13 min, p =0.04, Table 1); however, the mean difference between the two groups was less than 10 min.
Fig. 2 ROI methodology for quantitative data extraction. a Arterial ROIs. Left: For each PET/CT slice, a fitted arterial ROI was drawn manually to extract one SUVmax per slice (one ROI
corresponds to one slice). The compilation of all slices permitted the extraction of the highest and the average arterial SUVmax for
the three arterial territories (ascending aorta, aortic arch and descending aorta). Right: Images of fused coronal aortic PET/CT slices, and images of fused and nonfused axial aortic PET/CT slices of a GCA patient and a negative control. b Background ROIs. Upper left: Venous blood pool: eight contiguous fitted ROIs were manually drawn on the right jugular internal vein. From these, the highest and average venous SUVmax for
each patient were determined. Bottom left: Lung: a 2-cm circular ROI was drawn for each patient in a low-activity lung region far from any suspiciously noisy locoregional activity. From this, one lung SUVmax per patient was obtained. Right: Liver: for each patient, a fitted ROI was drawn on each liver slice to extract one SUVmax per slice. The
compilation of all liver slices permitted the determination of the highest and the average liver SUVmax for each patient
Eur J Nucl Med Mol Imaging (2014) 41:155166 159
Table 1 Characteristics of included patients
GCA (n =11) Controls (n =11) p value
Age (years) 73.65.63 734.72 Matched Sex
Male 4 3 Matched Female 7 8
Disease All TAB-proven GCA Neoplasia (8) IgG disease (1)
Focal lung infection (1) Suprarenal nodule (1)
CRP (mg/L) 17.924.1 39.150.5 0.28 Glycaemia (g/L) 1.00.23 1.00.18 0.58 Weight (kg) 68.313.78 60.515.76 0.25 Activity injected (MBq) 262.861.3 257.264.7 0.83 Delay time (min)a 68.5510.96 60.364.13 0.04 CT dose index (mGy) 5.931.42 5.371.46 0.28 Doselength product (mGy.cm) 539.46136.97 437.73149.62 0.21 Atheroma grade (CT)b
0 1 1
1 2 2 2 8 7 3 0 1 Number of arterial ROIs
Ascending aorta 164 161 0.81 Aortic arch 125 113 0.53 Descending aorta 516 476 0.18 Number of background ROIs
Liver 527 507 0.59Lung 1 1 Not significant Venous blood pool 8 8 Not significant
a In 5 of 11 GCA patients the delay time before acquisition was more than 67 min, which led to a significant difference between the two groups. However, the mean difference between the two groups was less than 10 min.
b According to the classification described by Tatsumi et al. [27]
Semiquantitative analysis
There were no significant differences in the mean numbers of generated ROIs per patient in the two groups for arterial territories or backgrounds (Table 1). For the three arterial territories (ascending aorta, arch and descending aorta), the arterial SUVmax values within the entire thoracic aorta were significantly higher in the GCA group than in the control group (Fig. 3). The greatest differences were observed in the aortic arch and descending aorta. Regarding the different background activities, the liver SUVmax values were significantly higher in the GCA group, in contrast to the lung and right jugular venous SUVmax values, in which no differences were observed (Fig. 4). Of the 22 patients, 20 (10 GCA and 10 controls, 91 % of the study population) showed calcifications in at least one thoracic aortic territory. Four patients (2 GCA and 2 controls) had grade I calcification, 15 patients (8 GCA and 7 controls) had grade II calcification, and 1 control had grade III calcification (Table 1). All matched pairs showed concordant calcification grading in at least one arterial territory. Of the 11 matched pairs, 7 were graded similarly for one arterial territory
and 4 for two arterial territories. The overall calcification grading was not significantly different between the two groups.
The type of background (liver, lung or venous blood pool) that was used to normalize the arterial thoracic activity and the arterial territory significantly affected discrimination between GCA patients and controls. None of the methods that were applied to the ascending aorta allowed discrimination between controls and GCA patients (Table 2, Fig. 5). When applied to the aortic arch, methods A, B, C and C provided a significant difference between the two groups (Table 2, Fig. 5). When applied both to the aortic arch and the descending aorta, normalization to the venous blood pool was able to significantly discriminate between GCA patients and controls (Table 2, Fig. 5c) and provided the best diagnostic performance (Fig. 6). In particular, method C applied to the aortic arch provided optimal diagnostic performance (sensitivity 81.8 %, specificity 91 %, AUC 0.87; p <0.0001) for a cut-off value of 1.53 (Fig. 6b).
CRP levels in the two groups were not significantly correlated with arterial uptake in any arterial territory, with the exception of the ascending aorta in the GCA group (r =0.67, p =0.03, confidence interval 0.10; 0.91). Nevertheless, the wide range of the
160 Eur J Nucl Med Mol Imaging (2014) 41:155166
Fig. 3 Arterial activities, per group and per arterial territory. For each arterial territory, SUVmax values in the GCA group were frequently higher than those in the control group (p <0.0001). Nevertheless, a high
frequency of overlap was observed for SUVmax values ranging from 2 to 4 in the three territories
confidence intervals (0.10 to 0.90) did not favour a consistent correlation (see Fig. 7 in the electronic supplementary material).
Discussion
This is the first study in which 18F-FDG PET/CT outcomes according to multiple currently available semiquantitative methods were compared in the assessment of aortic inflammation in GCA. A recent systematic review and meta-analysis reported on the heterogeneity in the available qualitative and semiquantitative methods and highlighted the diagnostic performance of 18F-FDG PET and PET/CT semiquantitative ratios [25]. In this study, the two most powerful semiquantitative arterial to background approaches that had been previously tested in GCA were compared with a third method that had been applied in atheroma but not in GCA [26]. Discrimination between the two groups based on normalization of the arterial activity to different backgrounds (liver, lung or blood pool) depended on the arterial territory and the background used. Although higher arterial activity in the GCA group than in the control group was observed in all three arterial territories, normalizations based on the venous blood pool provided the best discrimination.
Utility of SUVmax and background activity
As the GCA patients showed higher arterial activities than the controls in the three arterial territories, it might seem easier to use
semiquantitative SUVmax analysis directly. However, several reports contradict this approach. In a retrospective controlled
PET study, Lehmann et al. found that a secondary SUVmax
analysis compared to the initial visual analysis increased the detection sensitivity (90 % vs. 65 %) but significantly decreased the specificity (45 % versus 80 %) for detecting arterial wall inflammation in large vessel vasculitis [22]. Our matched-pairs controlled analysis showed frequently overlapping values in the three thoracic arterial territories between the two groups. Regarding the distribution of the SUVmax in the two groups
(Fig. 3), it appears that an optimal SUVmax cut-off value cannot be defined. An arbitrary cut-off would undoubtedly be dependent on the selected population and should not be extrapolated from the general population. Other methods in which the arterial SUVmax is normalized to the background activity seem to outperform the qualitative and SUVmax
approaches [2325]. In this study the degree of discrimination between the GCA and control groups depended on the chosen background (Table 2, Figs. 4 and 5).
Our results suggest that the liver background may not be as relevant as was suggested by Hautzel et al. [24]. The liver SUVmax differed significantly between our two matched-pairs groups, and the values in the GCA group were significantly higher than in the control group. Additionally, we observed a wide intragroup variability (Fig. 4a). Systemic inflammation in GCA affects liver metabolism and can influence the calculation of liver uptake ratios. Other limitations include regional physiological activities such as those in the urinary tract and hepatic hilum, which can lead to errors in the accurate
Eur J Nucl Med Mol Imaging (2014) 41:155166 161
Fig. 4 Background activity in each group. a Liver. SUVmax values were significantly different between the groups (p <0.0001), with higher values in the GCA group. SUVmax values ranged widely within the two groups (from 1.41 to 5.51, with a mean of 3.35, in the control group and from1.55 to 5.96, with a mean of 3.79, in the GCA group). b Lung. SUVmax
values were not significantly different between the groups (p =0.28). In the control group, SUVmax ranged from 0.29 to 0.70, with a mean of 0.44.
In the GCA group, SUVmax ranged from 0.31 to 0.62, with a mean of0.46. The between-group difference was 0.020.15. c Venous blood pool. SUVmax values were not significantly different between the groups (p =0.17). In the control group, SUVmax ranged from 0.97 to 2.66, with a mean of 1.75. In the GCA group, SUVmax ranged from 1.11 to 2.71, with a mean of 1.82. The between-group difference was 0.0760.55
delineation of ROIs. These limitations work against the use of the liver as the background for normalization in GCA. Moosig et al. proposed normalization to the lung because parenchymal lung activity shows low interpatient and intrapatient variability due to low physiological uptake of 18F-FDG [23].
In this study, the SUVmax did not differ significantly
between the two groups, and the interpatient variability was very low (Fig. 4b). Method B, when applied to the aortic arch,
provided good diagnostic performance with an optimal cut-off value of 7.46 (sensitivity 81.8 %, specificity 72.7 %, AUC0.79; p =0.003; Fig. 6b). Nevertheless, the same method applied to the other arterial territories, as well as method B applied to all arterial territories, did not provide significant differences (Table 2, Figs. 5b and 6). The arterial to venous blood pool uptake ratio is highly reproducible when applied in atheroma for the evaluation of arterial inflammation [26].
Table 2 Comparison of all methods by arterial territory
Method SUVmax Normalized to Uptake ratio (control vs. GCA)
Ascending aorta Aortic arch Descending aorta
A Highest Liver 0.760.07 vs. 0.790.15 0.730.06 vs. 0.870.24 0.810.12 vs. 0.930.31 A Average Liver 0.790.04 vs. 0.830.13 0.810.04 vs. 0.930.2* 0.770.05 vs. 0.870.22 B Highest Lung 7.631.93 vs. 8.752.4 7.191.63 vs. 9.52.5* 8.121.82 vs. 10.113.04 B Average Lung 6.421.55 vs. 7.211.91 6.51.45 vs. 8.02.2 6.171.26 vs. 7.52.2C Highest Blood poola 1.510.27 vs. 1.750.42 1.420.20 vs. 1.910.53** 1.620.31 vs. 2.030.63* C Average Blood poola 1.560.33 vs. 1.750.37 1.590.33 vs. 1.950.43* 1.510.30 vs. 1.830.45*
*p <0.05, **p <0.005
a Right internal jugular vein
162 Eur J Nucl Med Mol Imaging (2014) 41:155166
Our results showed no significant difference between the two groups and low interpatient variability (Fig. 4c). Normalization by blood pool activity outperformed the other methods in our study and permitted the differentiation of GCA from control patients when applied to the aortic arch and descending aorta (Figs. 5 and 6, Table 2). Method C, when applied to the aortic arch, provided the best overall performance for an optimal cut-off value of 1.53 (sensitivity81.8 %, specificity 91 %, AUC 0.87; p <0.0001; Fig. 6b). The venous activity reflects the blood pool, and normalization to this background may provide a good approximation of arterial wall activity.
Do calcification, sex and age affect arterial 18F-FDG uptake?
18F-FDG is not taken up by the normal aortic arterial wall but is trapped primarily by macrophages in atherosclerotic
Fig. 6 ROC analyses for each arterial territory: diagnostic performance. a Ascending aorta. b Aortic arch. c Descending aorta. ROC curves for each method applied to each arterial territory are shown. Sensitivities, specificities and AUC with their corresponding p values are also provided. Method C, when applied to the aortic arch, provided the best sensitivity and specificity (81.8 % and 91 %, respectively), with the highest AUC (0.87) and the most significant p value (p <0.0001), for a cut-off value of 1.53
Fig. 5 Arterial to background uptake ratios: between-group comparisons for each arterial territory. a Arterial to liver ratios. b Arterial to lung ratios. c Arterial to blood pool ratios. The values (y axis) of arterial to
background ratios in GCA patients (red) and their matched controls (blue) are compared for each vascular territory (x axis, ascending aorta, aortic arch and descending aorta, respectively). *p <0.05. **p <0.005
Eur J Nucl Med Mol Imaging (2014) 41:155166 163
164 Eur J Nucl Med Mol Imaging (2014) 41:155166
plaques [29, 30]. Several studies have highlighted the potential impact of different factors, including age, sex and calcification, on arterial wall 18F-FDG uptake [27, 3133]. As in previous studies by Tatsumi et al. [27] and Ben-Haim et al. [31], arterial calcifications and 18F-FDG uptake rarely overlapped. This result suggests that atherosclerosis is a global disease with different lesional phases and demonstrates that 18F-FDG uptake reveals active focal noncalcified lesions [27, 31, 32]. We matched controls and GCA patients using age and sex to highlight the potential power of our tested methods to discriminate between the two groups by reducing the suspected impact of these parameters on arterial 18F-FDG uptake. Additionally, we found calcifications in at least one region of the thoracic aortic wall in 91 % of included patients, with all matched pairs showing concordant calcification grading in at least one arterial territory. As the overall calcification grading was not significantly different between the two groups, the differences in arterial wall activity were independent of calcification grade, age and sex.
By analysing the thickness and the contrast enhancement of the arterial wall, contrast-enhanced CT provides good performance [13]. Wall thickness can persist long after the acute phase of arterial inflammation, limiting its relevance in the early assessment of inflammation. Arterial wall enhancement distinguishes active vasculitis from fibrosis. Combining the metabolic information obtained from 18F-FDG PET with the wall enhancement and thickness assessed by contrast-enhanced CT (in other words, contrast-enhanced FDG PET/CT) could be a powerful approach in the evaluation of extracranial GCA. Further studies should be performed to evaluate the usefulness of this approach in GCA.
Potential limitations
Potential limitations of our study were its retrospective design and small sample size. Nevertheless, our methodology was based on a controlled pair-matched design, anonymization of the included patients and randomization of patient image analysis, all of which limited the risks of measurement and confounding biases.
The diagnosis of GCAwas based on the 1990 ACR criteria, which include clinical, biological and histological items [5]. The presence of three or more ACR criteria allows a diagnosis with greater than 90 % sensitivity and specificity, but up to 16 % of GCA patients have atypical symptoms [4, 15]. Despite a high false-negative rate, TAB remains the gold standard for diagnosing GCA in daily practice. All of our GCA patients were positive for TAB. Additional biopsies of the aorta were not performed. We assumed that some of our GCA patients may not have had aortic involvement associated with their biopsy-proven cranial involvement. However, extracephalic GCA may occur in 6080 % of patients with cranial GCA [14, 34]. Furthermore, our GCA group had
significantly higher thoracic aortic uptake than our control group (Fig. 3). This difference could not be explained by age, sex or calcification status, as all of our GCA patients and controls were matched for age and sex, with similar arterial calcification status (Table 2).
We focused our analysis on the thoracic aorta, which provides high positive and negative predictive values for GCA diagnosis when assessed by 18F-FDG PET [35]. Subclavian arteries are frequently involved in GCA [25], but manual ROIs are approximate on non-contrast-enhanced PET/CT images and may overlap with anatomical regions, especially when arterial activity is low. If subclavian arteries are accurately localized, partial-volume effects may affect SUVmax measurements and corresponding ratio calculations. The abdominal aorta is less frequently involved in GCA than the thoracic aortic subregions [25, 36], and for this reason, we excluded this arterial territory from our analysis.
The majority of our GCA patients (73 %) were receiving corticosteroids when they underwent 18F-FDG PET/CT. However, Blockmans et al. reported that despite a clinical or biological response under corticosteroid treatment, GCA patients have persistent arterial wall inflammation that can be assessed by FDG PET [34]. Although the between-group differences in arterial uptake may have been underestimated due to the immunosuppression in a proportion of the GCA patients, the normalization of arterial wall activity to the blood pool background in our study still discriminated the two groups better than liver or lung normalization (Fig. 6). If our results are confirmed by further prospective studies, they could lead to a more accurate assessment of treatment responses and patient management.
The delays before image acquisition were slightly different between the two groups being longer in the GCA group, and this could have been a limitation of this study. This limitation was the same for all of the methods evaluated and for all arterial territories. Moreover, methods C and C, when applied to the ascending aorta, were not statistically relevant compared to the aortic arch and descending aorta (Fig. 6). In addition, the blood pool activity in the five GCA patients with a longer delay time was not significantly different from that in their matched paired controls or from that in the six GCA patients with a normal delay time for either SUVmax or SUVmean (p >
0.05). In addition, the arterial activities were not significantly different between the GCA group with the longer delay time and the GCA group with the normal delay time in any of the arterial territories (p >0.05 for both SUVmax and SUVmean).
Consequently, the 10-min longer delay time in the five GCA patients did not explain the difference in uptake ratios between the two groups in this study.
Conclusion
Our results suggest that among all of the semiquantitative methods based on 18F-FDG PET in GCA, the aortic to blood
Eur J Nucl Med Mol Imaging (2014) 41:155166 165
pool uptake ratios outperform liver and lung ratios in the assessment of arterial inflammation. Blood pool ratios should be validated in further prospective studies and may become a gold standard in the assessment of aortic wall inflammation in GCA patients.
Conflicts of interest None.
References
1. Gonzalez-Gay MA, Vazquez-Rodriguez TR, Lopez-Diaz MJ, Miranda-Filloy JA, Gonzalez-Juanatey C, Martin J, et al. Epidemiology of giant cell arteritis and polymyalgia rheumatica. Arthritis Rheum. 2009;61: 145461. doi:http://dx.doi.org/10.1002/art.24459
Web End =10.1002/art.24459 .
2. Weyand CM, Goronzy JJ. Giant-cell arteritis and polymyalgia rheumatica. Ann Intern Med. 2003;139:50515.
3. Weyand CM, Goronzy JJ. Medium- and large-vessel vasculitis. N Engl J Med. 2003;349:1609. doi:http://dx.doi.org/10.1056/NEJMra022694
Web End =10.1056/NEJMra022694 .
4. Salvarani C, Cantini F, Hunder GG. Polymyalgia rheumatica and giant-cell arteritis. Lancet. 2008;372:23445. doi:http://dx.doi.org/10.1016/S0140-6736(08)61077-6
Web End =10.1016/S0140- http://dx.doi.org/10.1016/S0140-6736(08)61077-6
Web End =6736(08)61077-6 .
5. Hunder GG, Bloch DA, Michel BA, Stevens MB, Arend WP, Calabrese LH, et al. The American College of Rheumatology 1990 criteria for the classification of giant cell arteritis. Arthritis Rheum. 1990;33:11228.
6. Hauser WA, Ferguson RH, Holley KE, Kurland LT. Temporal arteritis in Rochester, Minnesota, 1951 to 1967. Mayo Clin Proc. 1971;46:597602.
7. Hall S, Persellin S, Lie JT, OBrien PC, Kurland LT, Hunder GG. The therapeutic impact of temporal artery biopsy. Lancet. 1983;2:121720.8. Roth AM, Milsow L, Keltner JL. The ultimate diagnoses of patients undergoing temporal artery biopsies. Arch Ophthalmol. 1984;102: 9013.
9. Gonzalez-Gay MA, Garcia-Porrua C, Llorca J, Gonzalez-Louzao C, Rodriguez-Ledo P. Biopsy-negative giant cell arteritis: clinical spectrum and predictive factors for positive temporal artery biopsy. Semin Arthritis Rheum. 2001;30:24956. doi:http://dx.doi.org/10.1053/sarh.2001.16650
Web End =10.1053/sarh.2001.16650 .
10. Gonzalez-Gay MA, Garcia-Porrua C, Pineiro A, Pego-Reigosa R, Llorca J, Hunder GG. Aortic aneurysm and dissection in patients with biopsy-proven giant cell arteritis from northwestern Spain: a population-based study. Medicine (Baltimore). 2004;83:33541.
11. Bongartz T, Matteson EL. Large-vessel involvement in giant cell arteritis. Curr Opin Rheumatol. 2006;18:107.
12. Schmidt WA, Seifert A, Gromnica-Ihle E, Krause A, Natusch A. Ultrasound of proximal upper extremity arteries to increase the diagnostic yield in large-vessel giant cell arteritis. Rheumatology (Oxford). 2008;47:96101. doi:http://dx.doi.org/10.1093/rheumatology/kem322
Web End =10.1093/rheumatology/kem322 .
13. Agard C, Barrier JH, Dupas B, Ponge T, Mahr A, Fradet G, et al. Aortic involvement in recent-onset giant cell (temporal) arteritis: a case-control prospective study using helical aortic computed tomodensitometric scan. Arthritis Rheum. 2008;59:6706. doi:http://dx.doi.org/10.1002/art.23577
Web End =10. http://dx.doi.org/10.1002/art.23577
Web End =1002/art.23577 .
14. Brack A, Martinez-Taboada V, Stanson A, Goronzy JJ, Weyand CM. Disease pattern in cranial and large-vessel giant cell arteritis. Arthritis Rheum. 1999;42:3117. doi:http://dx.doi.org/10.1002/1529-0131(199902)42:2<311::AID-ANR14>3.0.CO;2-F
Web End =10.1002/1529-0131(199902) http://dx.doi.org/10.1002/1529-0131(199902)42:2<311::AID-ANR14>3.0.CO;2-F
Web End =42:2<311::AID-ANR14>3.0.CO;2-F .
15. Calamia KT, Hunder GG. Giant cell arteritis (temporal arteritis) presenting as fever of undetermined origin. Arthritis Rheum. 1981;24: 14148.
16. Warburg O. On the origin of cancer cells. Science. 1956;123:30914.17. Gillies RJ, Robey I, Gatenby RA. Causes and consequences of increased glucose metabolism of cancers. J Nucl Med. 2008;49 Suppl 2:24S42S. doi:http://dx.doi.org/10.2967/jnumed.107.047258
Web End =10.2967/jnumed.107.047258 .
18. Vander Heiden MG, Cantley LC, Thompson CB. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science. 2009;324:102933. doi:http://dx.doi.org/10.1126/science.1160809
Web End =10.1126/science.1160809 .
19. Blockmans D, Maes A, Stroobants S, Nuyts J, Bormans G, Knockaert D, et al. New arguments for a vasculitic nature of polymyalgia rheumatica using positron emission tomography. Rheumatology (Oxford). 1999;38:4447.
20. Meller J, Strutz F, Siefker U, Scheel A, Sahlmann CO, Lehmann K, et al. Early diagnosis and follow-up of aortitis with [18F]FDG PET and MRI. Eur J Nucl Med Mol Imaging. 2003;30:7306. doi:http://dx.doi.org/10.1007/s00259-003-1144-y
Web End =10. http://dx.doi.org/10.1007/s00259-003-1144-y
Web End =1007/s00259-003-1144-y .
21. Henes JC, Muller M, Krieger J, Balletshofer B, Pfannenberg AC, Kanz L, et al. [18F] FDG-PET/CT as a new and sensitive imaging method for the diagnosis of large vessel vasculitis. Clin Exp Rheumatol. 2008;26:S4752.
22. Lehmann P, Buchtala S, Achajew N, Haerle P, Ehrenstein B, Lighvani H, et al. 18F-FDG PET as a diagnostic procedure in large vessel vasculitis a controlled, blinded re-examination of routine PET scans. Clin Rheumatol. 2011;30:3742. doi:http://dx.doi.org/10.1007/s10067-010-1598-9
Web End =10.1007/s10067-010-1598-9 .
23. Moosig F, Czech N, Mehl C, Henze E, Zeuner RA, Kneba M, et al. Correlation between 18-fluorodeoxyglucose accumulation in large vessels and serological markers of inflammation in polymyalgia rheumatica: a quantitative PET study. Ann Rheum Dis. 2004;63: 8703. doi:http://dx.doi.org/10.1136/ard.2003.011692
Web End =10.1136/ard.2003.011692 .
24. Hautzel H, Sander O, Heinzel A, Schneider M, Muller HW. Assessment of large-vessel involvement in giant cell arteritis with 18F-FDG PET: introducing an ROC-analysis-based cutoff ratio. J Nucl Med. 2008;49:110713. doi:http://dx.doi.org/10.2967/jnumed.108.051920
Web End =10.2967/jnumed.108.051920 .
25. Besson FL, Parienti JJ, Bienvenu B, Prior JO, Costo S, Bouvard G, et al. Diagnostic performance of 18F-fluorodeoxyglucose positron emission tomography in giant cell arteritis: a systematic review and meta-analysis. Eur J Nucl Med Mol Imaging. 2011;38:176472. doi:http://dx.doi.org/10.1007/s00259-011-1830-0
Web End =10.1007/s00259-011-1830-0 .26. Rudd JH, Myers KS, Bansilal S, Machac J, Pinto CA, Tong C, et al. Atherosclerosis inflammation imaging with 18F-FDG PET: carotid, iliac, and femoral uptake reproducibility, quantification methods, and recommendations. J Nucl Med. 2008;49:8718. doi:http://dx.doi.org/10.2967/jnumed.107.050294
Web End =10.2967/ http://dx.doi.org/10.2967/jnumed.107.050294
Web End =jnumed.107.050294 .
27. Tatsumi M, Cohade C, Nakamoto Y, Wahl RL. Fluorodeoxyglucose uptake in the aortic wall at PET/CT: possible finding for active atherosclerosis. Radiology. 2003;229:8317. doi:http://dx.doi.org/10.1148/radiol.2293021168
Web End =10.1148/radiol. http://dx.doi.org/10.1148/radiol.2293021168
Web End =2293021168 .
28. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte Jr M, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol. 1990;15: 82732.
29. Rudd JH, Warburton EA, Fryer TD, Jones HA, Clark JC, Antoun N, et al. Imaging atherosclerotic plaque inflammation with [18F]-fluorodeoxyglucose positron emission tomography. Circulation. 2002;105:270811.
30. Ogawa M, Ishino S, Mukai T, Asano D, Teramoto N, Watabe H, et al. (18)F-FDG accumulation in atherosclerotic plaques: immunohistochemical and PET imaging study. J Nucl Med. 2004;45: 124550.
31. Ben-Haim S, Kupzov E, Tamir A, Israel O. Evaluation of 18F-FDG uptake and arterial wall calcifications using 18F-FDG PET/CT. J Nucl Med. 2004;45:181621.
32. Dunphy MP, Freiman A, Larson SM, Strauss HW. Association of vascular 18F-FDG uptake with vascular calcification. J Nucl Med. 2005;46:127884.
33. Bural GG, Torigian DA, Chamroonrat W, Houseni M, Chen W, Basu S, et al. FDG-PET is an effective imaging modality to detect and
166 Eur J Nucl Med Mol Imaging (2014) 41:155166
quantify age-related atherosclerosis in large arteries. Eur J Nucl Med Mol Imaging. 2008;35:5629. doi:http://dx.doi.org/10.1007/s00259-007-0528-9
Web End =10.1007/s00259-007-0528-9 .34. Blockmans D, de Ceuninck L, Vanderschueren S, Knockaert D, Mortelmans L, Bobbaers H. Repetitive 18F-fluorodeoxyglucose positron emission tomography in giant cell arteritis: a prospective study of 35 patients. Arthritis Rheum. 2006;55:1317. doi:http://dx.doi.org/10.1002/art.21699
Web End =10.1002/ http://dx.doi.org/10.1002/art.21699
Web End =art.21699 .
35. Blockmans D, Stroobants S, Maes A, Mortelmans L. Positron emission tomography in giant cell arteritis and polymyalgia rheumatica: evidence for inflammation of the aortic arch. Am J Med. 2000;108: 2469.
36. Lie JT. Aortic and extracranial large vessel giant cell arteritis: a review of 72 cases with histopathologic documentation. Semin Arthritis Rheum. 1995;24:42231.
Springer-Verlag Berlin Heidelberg 2014