European Journal of Nuclear Medicine and Molecular Imaging

, Volume 38, Issue 9, pp 1764–1772

Diagnostic performance of 18F-fluorodeoxyglucose positron emission tomography in giant cell arteritis: a systematic review and meta-analysis

Authors

  • Florent L. Besson
    • Department of Nuclear MedicineCHU Caen, Université Caen Basse-Normandie
  • Jean-Jacques Parienti
    • Department of BiostatisticsCHU Caen, Université Caen Basse-Normandie
  • Boris Bienvenu
    • Department of Internal MedicineCHU Caen, Université Caen Basse-Normandie
  • John O. Prior
    • Department of Nuclear MedicineCentre Hospitalier Universitaire Vaudois
  • Sylvie Costo
    • Department of Nuclear MedicineCHU Caen, Université Caen Basse-Normandie
  • Gerard Bouvard
    • Department of Nuclear MedicineCHU Caen, Université Caen Basse-Normandie
    • Department of Nuclear Medicine, EA 3212 (Coeur et Ischémie)CHU Caen, Université Caen Basse-Normandie
Review Article

DOI: 10.1007/s00259-011-1830-0

Cite this article as:
Besson, F.L., Parienti, J., Bienvenu, B. et al. Eur J Nucl Med Mol Imaging (2011) 38: 1764. doi:10.1007/s00259-011-1830-0
  • 697 Views

Abstract

Purpose

The aim of this study was to conduct a systematic review and perform a meta-analysis on the diagnostic performances of 18F-fluorodeoxyglucose positron emission tomography (FDG PET) for giant cell arteritis (GCA), with or without polymyalgia rheumatica (PMR).

Methods

MEDLINE, Embase and the Cochrane Library were searched for articles in English that evaluated FDG PET in GCA or PMR. All complete studies were reviewed and qualitatively analysed. Studies that fulfilled the three following criteria were included in a meta-analysis: (1) FDG PET used as a diagnostic tool for GCA and PMR; (2) American College of Rheumatology and Healey criteria used as the reference standard for the diagnosis of GCA and PMR, respectively; and (3) the use of a control group.

Results

We found 14 complete articles. A smooth linear or long segmental pattern of FDG uptake in the aorta and its main branches seems to be a characteristic pattern of GCA. Vessel uptake that was superior to liver uptake was considered an efficient marker for vasculitis. The meta-analysis of six selected studies (101 vasculitis and 182 controls) provided the following results: sensitivity 0.80 [95% confidence interval (CI) 0.63–0.91], specificity 0.89 (95% CI 0.78–0.94), positive predictive value 0.85 (95% CI 0.62–0.95), negative predictive value 0.88 (95% CI 0.72–0.95), positive likelihood ratio 6.73 (95% CI 3.55–12.77), negative likelihood ratio 0.25 (95% CI 0.13–0.46) and accuracy 0.84 (95% CI 0.76–0.90).

Conclusion

We found overall valuable diagnostic performances for FDG PET against reference criteria. Standardized FDG uptake criteria are needed to optimize these diagnostic performances.

Keywords

PET18F-FDGGiant cell arteritisPolymyalgia rheumatica

Introduction

Giant cell arteritis (GCA) is the most common form of vasculitis in Western countries. The incidence of 20 per 100,000 people older than 50 years of age (female to male ratio, 2:1) based on autopsy studies is probably an underestimation [1]. Initially described as temporal arteritis (Horton disease) [2], GCA is a segmental panarteritis that involves intracranial or extracranial vessels. The thoracic aorta and its main branches are involved in 45% of newly diagnosed patients [3]. At diagnosis, short-term prognosis is driven by ophthalmic complications that require urgent corticosteroid treatment to prevent the development of definitive blindness [46]. Long-term prognosis is determined by the presence of thoracic aortic aneurysms and stenosis [7, 8]. Polymyalgia rheumatica (PMR) is a common painful rheumatological syndrome that is three times more frequent than GCA [912] and develops prior to, concurrently with or after GCA. In general, 16–21% of PMR patients have GCA, and more than 50% of GCA cases present with PMR symptoms [11, 1315]. The aetiology of both disorders remains unclear.

Due to the close relationship between these disorders and the common major histocompatibility complex profile, the scientific community now considers them as polygenic entities of a single general disease [16]. Nevertheless, the diagnostic criteria remain empirical and independent for both entities, which result in frequent misclassification with the risk of suboptimal therapy management [1719]. A temporal artery biopsy, which is the historical diagnostic gold standard for cranial GCA, provides high false-negative rates (10–40%) [2024] and is unsuitable for the diagnosis of extracranial arteritis. Thus, the general American College of Rheumatology criteria (including temporal artery biopsy) are used as the reference standard in clinical practice [25]. Three of five positive criteria provide more than 90% sensitivity and specificity [25], but nonspecific symptoms, in up to 16% of cases, such as fever of unknown origin, remain a clinical issue [26, 27]. For PMR, the use of Healey diagnostic reference criteria [28] can lead to the misdiagnosis of subclinical GCA [17, 19]. The lack of an efficient adapted noninvasive diagnostic tool has promoted the use of imaging modalities.

Ultrasonography is cost-effective and very sensitive for the detection of cranial vasculitis (characteristic “halo sign”), but its inability to correctly evaluate thoracic arteries and its high operator dependency has limited its role in extracranial cases [29, 30]. MRI provides high-resolution images of anatomical vessel walls. Increased wall thickness, oedema and mural contrast enhancement suggest nonspecific vessel wall inflammation [31]. Although useful to detect temporal arteritis [30], MRI needs further evaluation before it can be routinely used to determine extracranial GCA.

18F-fluorodeoxyglucose positron emission tomography (FDG PET) is widely used in oncology. Its application in vasculitis was first described by Blockmans et al. in 1999 [32]. Since 2000, several studies have evaluated its utility to detect vessel wall inflammation in arteritis; this has influenced the re-evaluation of PMR as a particular form of GCA. The aim of this study was to (1) collect the data from the literature and (2) perform a meta-analysis of the performances of FDG PET to detect GCA, overlapped or not with PMR.

Materials and methods

Database search

MEDLINE, Embase and the Cochrane Library were searched for articles written in English, up to November 2011, that addressed FDG PET as a diagnostic tool used in cases of GCA or PMR. We used the MeSH query “giant cell arteritis” or “polymyalgia rheumatica” and “positron emission tomography”. An initial selection was based on the exclusion of abstracts and case reports because they failed to provide sufficient data for analysis. A second step consisted of selecting original complete studies with a consistent number of patients (superior or equal to eight). To assess the diagnostic performance of FDG PET, comparable studies that fulfilled all three of the following criteria were included in the meta-analysis: (1) FDG PET used as a diagnostic tool to determine GCA or PMR; (2) American College of Rheumatology and Healey criteria used as a reference standard for the diagnosis of GCA and PMR, respectively; and (3) the use of a control group.

Methodological assessment

For original articles, FDG interpretation criteria, uptake pattern, topography and the influence of biological markers [erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP)] were analysed. Based on this qualitative review, meta-analysis was performed of comparable studies that fulfilled all the inclusion criteria. Two independent reviewers evaluated the methodology of the selected studies, using the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS) [33, 34]. The evaluation was based on a 14-point scale (spectrum composition, selection criteria, reference standard, disease progression bias, partial verification, differential verification, incorporation bias, index test execution, reference standard execution, test review bias, reference standard review bias, clinical review bias, non-interpretable test results and explanation for withdrawal). A study with a score below 7 was considered to be of low quality, one with a score of 7–10 was considered to be of good quality and one with a score of 10–14 was considered to be of high quality. Reviewers, who were blinded to the purposes of the meta-analysis, recorded a score of “1” for “yes” and “0” for “no” for each of the 14 points on the scale, according to the QUADAS detail list. All disagreements were resolved by consensus.

Data extraction and statistical analysis

For the meta-analysis, the number of true-positives (TP), true-negatives (TN), false-positives (FP) and false-negatives (FN) were extracted or computed from each selected study based on the FDG PET as the index test, and the American College of Rheumatology or the Healey criteria as the reference diagnosis test. Then, the pooled sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratio and accuracy were calculated. The between-studies heterogeneity was assessed using the tau-squared and I-squared tests. The tau-squared test provided an estimate of the between-study variance, and the I-squared test quantified the extent of the heterogeneity (i.e. the percentage of total variation due to between-study variances). Instead of simple assessment being based on the inverse of the within-study variance, the weight was calculated using the inverse of the sum of the within-study and between-study variance estimates. To integrate the tau-squared weights, which accounted for between-study heterogeneities, we used a random-effects model. All statistical analyses were performed using MetaAnalyst® (Beta 3.13) [35].

Results

A total of 101 citations were found using the database searches. Thirteen complete and eligible studies were identified during the primary screening. We then extracted one additional full-text article through screening the studies and their references [36]. We first performed a qualitative analysis of the 14 complete articles and only included controlled studies in the meta-analysis to assess the performance of FDG PET to diagnose GCA with or without PMR (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00259-011-1830-0/MediaObjects/259_2011_1830_Fig1_HTML.gif
Fig. 1

Flowchart of the review process

Qualitative analysis: systematic review

Using the database search, 14 complete articles written over the past 10 years were found, including 12 prospective and 2 retrospective studies (Table 1).
Table 1

FDG PET as a diagnostic tool in GCA and PMR: systematic review

Author (reference)

Year

Design

Technique

FDG diagnostic criteria

Vasculitis

Blockmans et al. [32]

1999

Prospective

PET

Qualitative

11

Blockmans et al. [37]

2000

Prospective

PET

Qualitative

25

Meller et al. [38]

2003

Prospective

PET & PET/CT

Qualitative

15

Bleeker-Rovers et al. [39]

2003

Retrospective

PET

Qualitative

14

Moosig et al. [17]

2004

Prospective

PET

Qualitativea

13

Brodmann et al. [46]

2004

Prospective

PET

Not specified

22

Scheel et al. [36]

2004

Prospective

PET & PET/CT

Qualitative

8

Walter et al. [40]

2005

Prospective

PET

Qualitative

26

Blockmans et al. [44]

2006

Prospective

PET

Semi-quantitative

35

Blockmans et al. [18]

2007

Prospective

PET

Semi-quantitative

35

Henes et al. [42]

2008

Prospective

PET/CT

Qualitativea

13

Hautzel et al. [45]

2008

Prospective

PET

Semi-quantitative

18

Both et al. [41]

2008

Prospective

PET

Qualitative

25

Lehmann et al. [43]

2011

Retrospective

PET

Qualitativea

20

aFDG PET diagnostics were assessed by visual analysis and evaluated, secondarily, semi-quantitatively

Seven studies used exclusively qualitative FDG uptake criteria to diagnose vasculitis [32, 3641]. Blockmans et al.’s paper proposed a visual score interpretation that ranged from 0 (no uptake) to 3 (high uptake), using a score of 2–3 as positive [32, 37]. Both et al.’s study used this visual score without specifying a positive threshold [41]. Although first proposed by Meller et al. in 2003 [38], three other studies used a 4-point graded visual scale, based on the vessel to liver ratio. Based on this scale, 0 was defined as no uptake, 1 was defined as uptake less than that of the liver, 2 was defined as equal to liver uptake and 3 was defined as greater than liver uptake [36, 38, 40]. Two of these four studies concluded that grade 2–3 for the thoracic aorta, and a grade above 1 in other vascular regions, were positive criteria for vasculitis [38, 40]. One study described the vascular involvement of the thoracic aorta as “intense” for all vasculitides, without specifying the grade [36]. Bleeker-Rovers et al. based their interpretation exclusively on the topography (focal non-physiological accumulations of FDG as a positive criterion for vasculitis) [39].

Basing the diagnosis on visual uptake criteria, three other studies also performed secondary semi-quantitative analyses on vascular uptake [17, 42, 43]. In a prospective controlled study, Moosig et al. observed moderate to high vasculitis uptake in 12 of 13 polymyalgic patients. They found a mean semi-quantitative vessel to lung SUVmax ratio of 1.58 ± 0.37 [17]. Henes et al. found a mean maximum standardized uptake value (SUVmax) of 3.4 for all vasculitis cases (3.9 without corticosteroid treatment) [42]. Using receiver-operating characteristic analysis, Lehmann and coworkers retrospectively found an optimal sensitivity for the SUVmax cutoff point of 1.78 (90 vs 65% with visual assessment), though the corresponding specificity was decreased (45 vs 80% with visual assessment) [43].

Three studies evaluated semi-quantitative vessel uptake exclusively [18, 44, 45]. Hautzel et al. introduced a semi-quantitative aorta to liver SUVmax ratio in 18 GCA cases and 54 age- and sex-matched controls in a prospective study. Using receiver-operating characteristic analysis, the authors found optimal FDG PET overall performances for a cutoff ratio of 1.0 as a positive criterion for vasculitis [45]. Blockmans et al. developed a composite score for vascular uptake (total vascular score, TVS) in two non-controlled prospective studies that included 35 GCA cases with or without PMR [44], and 35 isolated cases of PMR [18]. For each patient, the TVS integrated vascular uptake (1, no uptake; 2, moderate uptake; and 3, high uptake) was found for seven predefined vascular regions (thoracic aorta, abdominal aorta, subclavian arteries, axillary arteries, carotid arteries, iliac arteries and femoral arteries). This method resulted in a score that ranged from 0 to 21. They found a mean TVS of 6 ± 0.2 at the time of diagnosis in 29 of 35 patients. This value was independent of the combined GCA-PMR condition [44]. They also found a mean TVS of 0.8 ± 1.7 in 31% of isolated PMR cases [18].

The main FDG PET-positive vascular territories observed in GCA included the thoracic aorta, the aortic arch [18, 32, 38, 4042, 44], the supra-aortic trunks, including the subclavian [17, 18, 32, 37, 38, 43, 44], and the carotid arteries [17, 38, 40]. Several studies have reported FDG uptake in the abdominal aorta [18, 38, 40, 42, 44] and the iliofemoral arteries [18, 38].

Brodmann et al. focused their analysis on the temporal arteries in a prospective study that included 22 GCA cases. All localized temporal arteritis cases that were positive, based on ultrasonography, were negative based on FDG PET. The authors concluded that FDG PET was not capable of diagnosing temporal arteritis due to the small diameter of the arteries and their proximity to the brain [46]. Blockmans et al. described a bilateral, symmetric, smooth and linear uptake pattern [32, 37], and semi-quantitatively confirmed their observations using TVS scores [44]. In terms of PMR, the authors observed a significant increase in FDG uptake in the shoulders, hips and vertebral spinous processes of GCA-PMR patients [18], and shoulder uptake correlated well with rheumatological symptoms (p = 0.005).

Of the 14 studies, 6 evaluated the relationship between FDG uptake at diagnosis and serological marker levels [17, 37, 4042, 44]. Four studies found no correlation between the intensity of FDG uptake or disease activity and the levels of either ESR or CRP [37, 41, 42, 44]. With high ESRmean and CRPmean values (84 ml/h and 106 mg/l, respectively), Moosig et al. found a significant correlation between both ESR or CRP levels and FDG vessel uptakes (r = 0.79, p < 0.0001 and r = 0.68, p < 0.001, respectively) [17]. Despite moderate biological inflammation (ESRmean = 35 ml/h and CRPmean = 27.4 mg/l) with a mean sensitivity of 57%, Walter et al. [40] found a significant correlation with the grade of uptake (ESR, p = 0.08; CRP, p = 0.002) and showed, through logistic regression, increased sensitivity of up to 90% for CRP levels superior to 160 mg/l.

Quantitative analysis: meta-analysis

Of the 14 complete studies, only 8 fulfilled all 3 inclusion criteria. Two studies were excluded because of the absence of a well-defined control design [32, 37]. Thus, we finally included six studies in the meta-analysis (Fig. 1). Based on the QUADAS criteria, three of the six studies were determined to be of high quality [38, 40, 43], two were determined to be of good quality [17, 45] and one was at the limit of good quality with a score of 7/14 [42] (Table 2). Pooled data included 81 GCA cases with or without PMR symptoms, 10 clinical PMR cases that fulfilled Healey criteria with subclinical GCA, as detected by FDG PET, and 182 controls. Positive temporal artery biopsies were noted in 17 of the 91 GCA-PMR cases included (19%). Qualitative and quantitative data were extracted from each selected study (Table 3). In Hautzel et al.’s study [45], data were computed from results provided by the authors (sensitivity, specificity and PPV based on an optimal aorta to liver ratio of 1.00), as follows: TP = sensitivity × (TP + FN), (TP + FN) representing the total number of true GCAs; TN = specificity × (TN + FP), (TN + FP) representing the total number of controls; FP = [TP × (1-PPV)]/PPV; and FN = total patients − (TP + TN + FP). In Meller et al.’s and Walter et al.’s studies, the same visual grading score was used [38, 40]. However, in Meller et al.’s study, the results between vasculitis and the controls were highly overlapped due to a lack of specificity, as pointed out by Hautzel et al. [45]. Furthermore, both studies proposed different positive grades depending on the localization of uptake. To homogenize quantitative data, and based on the results provided by Hautzel [45], we considered grade 2 (vascular uptake equal or superior to liver uptake) as a positive uptake criterion for vasculitis in both Meller et al.’s [38] and Walter et al.’s [40] studies, independent of vascular localization. In Moosig et al.’s study [17], 13 cases of vasculitis were mentioned, but only 12 FDG PETs were performed. Consequently, 12 vasculitis cases and 14 controls were evaluated. Data from the two other studies did not present any ambiguity [42, 43].
Table 2

QUADAS scores of the six controlled studies included in the meta-analysis

Study (reference)

Itemsa

Score

1

2

3

4

5

6

7

8

9

10

11

12

13

14

Meller et al. [38]

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

14

Moosig et al. [17]

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

No

No

Yes

Yes

Yes

12

Walter et al. [40]

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

14

Henes et al. [42]

Yes

No

No

Yes

No

No

Yes

Yes

No

No

No

Yes

Yes

Yes

7

Hautzel et al. [45]

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

No

No

Yes

Yes

Yes

12

Lehmann et al. [43]

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

14

a1 spectrum composition, 2 selection criteria, 3 reference standard, 4 disease progression bias, 5 partial verification, 6 differential verification, 7 incorporation bias, 8 index test execution, 9 reference standard execution, 10 test review bias, 11 reference standard review bias, 12 clinical review bias, 13 uninterpretable test results, 14 explanation for withdrawal

Table 3

Characteristics of the six controlled studies included in the meta-analysis

Author

(reference)

Year

Design

Technique

FDG diagnostic criteriaa

Treatmentb

Patient characteristicsc

Quantitative data

Vasculitis

Controls

Age

Sex ratio

Age

Sex ratio

TP

FN

TN

FP

n

Meller et al. [38]

2003

Prospective

PET & PET/CT

Visual grading scale

No

62

1.5

55

0.73

11

4

78

0

93

Moosig et al. [17]

2004

Prospective

PET

Visual uptake pattern

No

64

5

65

0.75

12

0

14

0

26

Walter et al. [40]

2005

Prospective

PET

Visual grading scale

Yes (65%)

71

4.2

71

4.2

15

11

26

0

52

Henes et al. [42]

2008

Prospective

PET/CT

Visual uptake pattern

Yes (70%)

65

4

57

3

9

1

8

0

18

Hautzel et al. [45]

2008

Prospective

PET

Aorta/liver SUVmax ratio

No

64

5

64

5

16

2

34

2

54

Lehmann et al. [43]

2011

Retrospective

PET

Visual uptake pattern

Yes (40%)

62

4

61

4

13

7

16

4

40

TP true-positive, FN false-negative, TN true-negative, FP false-positive, n total patients evaluated by PET or PET/CT

aMoosig: vessel/lung SUVmax secondary analysis, Henes and Lehmann: SUVmax secondary analysis

bTreatment at baseline PET or PET/CT, % of vasculitis undergoing corticosteroid treatment in parentheses

cAges are reported as mean age; sex ratio is calculated as the female to male ratio

FDG PET sensitivities ranged from 57 to 96% [17, 38], but only two of the six studies showed sensitivities under 85% (Fig. 2a) [40, 43]. All of the studies, except two, showed greater than 94% specificity (Fig. 2b) [38, 43]. FDG PET, compared to the reference clinical criteria, provided the following pooled performances: sensitivity 0.80 [95% confidence interval (CI) 0.63–0.91], specificity 0.89 (95% CI 0.78–0.94), PPV 0.85 (95% CI 0.62–0.95), NPV 0.88 (95% CI 0.72–0.95), positive likelihood ratio 6.73 (95% CI 3.55–12.77), negative likelihood ratio 0.25 (95% CI 0.13–0.46) and accuracy 0.84 (95% CI 0.76–0.90) (Table 4). I-squared results were statistically significant (p < 0.05) for all values except specificity, accuracy and likelihood ratios, justifying the use of a random-effects model.
https://static-content.springer.com/image/art%3A10.1007%2Fs00259-011-1830-0/MediaObjects/259_2011_1830_Fig2_HTML.gif
Fig. 2

a Forest plot of FDG PET sensitivity compared to the clinical reference standard. N total number of patients evaluated by PET or PET/CT. b Forest plot of FDG PET specificity compared to the clinical reference standard. N total number of patients evaluated by PET or PET/CT

Table 4

Pooled results: meta-analysis of the six controlled studies

Pooled performances

Value

95% confidence interval

Tau-squareda

I-squaredb

Lower

Upper

Sensitivity

0.80

0.63

0.91

0.61

66%

Specificity

0.89

0.78

0.94

0.31

50%

PPV

0.85

0.62

0.95

1.403

77%

NPV

0.88

0.72

0.95

1.026

77%

Accuracy

0.84

0.76

0.90

0.21

63%

Positive LR

6.73

3.55

12.77

0.19

46%

Negative LR

0.25

0.13

0.46

0.25

60%

PPV positive predictive value, NPV negative predictive value, LR likelihood ratio

aEstimate of the between-study variance

bQuantifies the extent of heterogeneity (i.e. the percentage of total variation due to between-study variances)

Discussion

To our knowledge, this meta-analysis is the first to evaluate global FDG PET diagnostic performance in cases of GCA.

The temporal artery biopsy criterion was mentioned as positive in 19% of the GCA-PMR cases for which meta-analysis was conducted. Based on these results, we consider FDG PET as a promising tool for the diagnostic assessment of GCA, particularly for extracranial GCA.

Our quantitative findings were consistently heterogeneous and could be explained in several ways. The methodological assessment of FDG uptake was not consistent across the studies. The vessel to liver visual grading score proposed by Meller et al. [38] provided high sensitivity (93%), but this reproducible visual evaluation lacks specificity, as does the SUVmax cutoff point proposed recently by Lehmann et al. [43]. However, the semi-quantitative aorta to liver ratio, proposed by Hautzel et al., provides optimal global performance and demonstrates robustness as an observer-independent method [45]. Another semi-quantitative vessel to lung ratio, proposed by Moosig et al., also provides good overall performance. Nevertheless, these authors initially based their diagnosis on a visual vessel uptake pattern and secondarily refined it to the qualitative criteria using semi-quantitative analysis [17]. A visual smooth linear or long segmental uptake pattern appears typical and should be considered as an important FDG PET criterion. Semi-quantitative analysis could improve overall performances and homogenize inter-observer interpretation. In this way, the aorta to liver or vessel to lung SUVmax ratios proposed by Hautzel et al. and Moosig et al. provided the most powerful results [17, 45].

Serological marker levels influenced by corticoid treatment could also explain part of the heterogeneity. As illustrated in Walter et al.’s study, high ESR and CRP levels significantly improved the sensitivity of FDG PET by forming a significant correlation between the grade of uptake and inflammation levels [40]. However, Walter et al.’s hypothesis contradicts that of Henes and coworkers, who had relatively similar treatment conditions (65 vs 70% of vasculitis cases were receiving corticoids, ESRmean = 35 vs 51 ml/h, CRPmean = 27.4 vs 38 mg/l, respectively) [42], and that of two of Blockmans et al.’s studies for which meta-analysis was not performed [37, 44]. A decrease in FDG uptake under treatment has been described during follow-up [17, 36, 38, 40], but disease activity or risk of relapse does not seem to be correlated with FDG or biological findings under therapy [18, 41, 44]. In comparison, biological findings are not consistent with disease activity in Takayasu arteritis [4749]. In the absence of standardized FDG PET interpretation criteria, the relationship between FDG uptake and serological inflammation levels remains ambiguous and should be interpreted with caution, especially in patients receiving corticoids.

Concerning index test modality, two of the six studies assessed by meta-analysis evaluated 18/101 vasculitis cases with PET/CT [38, 42]. Of potential interest could be the differentiation between vasculitis and atherosclerosis (such as calcification in the arterial wall in addition to focal, slight and moderate FDG uptake [50, 51]), or the detection of complications, such as stenoses or aneurysms from a single examination [42]; however, the potential added value of CT needs to be confirmed in further studies. The baseline differences among the patients in the studies may have contributed to the observed heterogeneity of the results too. However, such variability was accounted for in a random-effects model.

We tried not to include Takayasu arteritis in our analysis. Categorized as large-vessel vasculitis, GCA and Takayasu arteritis are two independent distinct diseases. Despite apparent similar FDG distributions, their pathophysiologies, evolution and prognoses are not comparable, nor are their respective population targets (age at diagnosis and sex ratio). Pooling both diseases to evaluate FDG PET diagnostic performances has limited clinical consistency. Focusing on the vascular component of GCA-PMR disease, we have tried to optimize the small number of studies amenable to meta-analysis while preserving the clinical coherence of FDG PET in GCA. We could exclude several Takayasu arteritis cases from the analysis [42], but ten Takayasu cases divided between three studies could not be separated out from the available data [38, 40, 43]. Our quantitative analysis included a total of 101 vasculitis cases (81 GCAs, with or without PMR symptoms, 10 clinical PMRs, with subclinical arteritis, 10 Takayasu cases) and 182 controls. For more clinical consistency, further studies should limit this population bias.

Subclinical GCA cases that do not fulfil the American College of Rheumatology criteria, but do meet the Healey criteria, remain a clinical issue due to the suboptimal nature of the actual reference standards used. Frequent resultant misclassification may suggest inappropriate management of patients and justifies the inclusion of PMR in our meta-analysis [1619]. The increased interest in FDG PET in GCA and PMR cases may contribute to redefining the actual diagnostic criteria.

In conclusion, FDG PET to detect extracranial GCA should be performed in patients with a negative temporal artery biopsy, isolated clinical PMR symptoms or atypical cases that do not fulfil the reference criteria. Our review highlights the need for standardized FDG diagnostic criteria to optimize the results.

Conflicts of interest

None.

Copyright information

© Springer-Verlag 2011