(18)F-FDG and (18)F-NaF PET/CT demonstrate coupling of inflammation and accelerated bone turnover in rheumatoid arthritis.

OBJECTIVE
To compare the findings in rheumatoid arthritis (RA)-affected joints between (18)F-fluorodeoxyglucose (FDG) and (18)F-fluoride (NaF) positron emission tomography (PET)/computed tomography (CT).


METHODS
We enrolled twelve RA patients who started a new biologic agent (naïve 9 and switch 3). At entry, both hands were examined by (18)F-FDG PET/CT, (18)F-NaF PET/CT, and X-ray. Intensity of PET signals was determined by standardized uptake value max (SUVmax) in metacarpophalangeal (MCP), proximal interphalangeal (PIP), and ulnar, medial, and radial regions of the wrists. Hand X-rays were evaluated according to the Genant-modified Sharp score at baseline and 6 months.


RESULTS
Both (18)F-FDG and (18)F-NaF accumulated in RA-affected joints. The SUVmax of (18)F-FDG correlated with that of (18)F-NaF in individual joints (r = 0.65), though detail distribution was different between two tracers. (18)F-NaF and (18)F-FDG signals were mainly located in the bone and the surrounding soft tissues, respectively. The sum of SUVmax of (18)F-NaF correlated with disease activity score in 28 joint (DAS28), modified health assessment questionnaire (MHAQ), and radiographic progression. (18)F-FDG and (18)F-NaF signals were associated with the presence of erosions, particularly progressive ones.


CONCLUSION
Our data show that both (18)F-FDG and (18)F-NaF PET signals were associated with RA-affected joints, especially those with ongoing erosive changes.


Introduction
Rheumatoid arthritis (RA) is a chronic infl ammatory disease characterized by autoimmunity and polyarticular synovial infl ammation; subsequently, there is destruction of cartilage and bone. The earliest bone change detected by X-ray is erosion, which is associated with progressive joint destruction in the late phase of RA, especially in patients not receiving appropriate therapy [1]. Positive autoantibodies such as anti-cyclic citrullinated peptide antibody (ACPA), rheumatoid factor (RF), and high disease activity are additional predisposing factors for joint destruction, indicating that autoimmune-mediated infl ammation is implicated in joint damage in both cartilage and bone [2 -5]. Specifi cally, in the " treatto-target " concept, management of RA primarily aims to induce early clinical remission and, subsequently, maintain responses that are mainly focused on suppressing infl ammation [6]. Prompt treatment reduces infl ammation, resulting in limited structural change and better long-term radiologic outcomes [7]. However, according According to recent studies, 18 F-FDG PET sensitively detects active infl ammation in rheumatic diseases, including synovitis in RA [12 -18]. In contrast, 18 F-NaF PET is an increasingly used molecular imaging modality in human skeletal disorders [19 -23]. 18 F-NaF incorporates into the bone at the site of bone formation or remodeling; there, osteoblasts and osteoclasts are activated, and 18 F ions exchange the hydroxyl ions in bone crystal to form fl uorapatite. In addition to bone metabolism, the ratelimiting step of 18 F-NaF bone uptake is blood fl ow [24]. 18 F-NaF PET is useful for monitoring pathologic changes in the bones of experimental arthritis [25] and illustrating osteoplastic lesions in human joint diseases, such as osteoarthritis (OA) [21] and psoriatic arthritis [22].
Here, we compared 18 F-FDG and 18 F-NaF PET/CT analyses of the hands with other imaging modalities and clinical assessment in RA patients requiring biologic disease-modifying antirheumatic drugs (bDMARDs) for active disease. Moreover, we investigated the relationship of these PET fi ndings with radiologic progression of the joint lesions.

Patients
We enrolled 12 patients (10 female, 2 male; average age, 60.0 Ϯ 15.9 years) at Yokohama City University Hospital who fulfi lled the American College of Rheumatology 1987 classifi cation criteria for RA [26]. Nine patients started the fi rst bDMARD (etanercept 4, infl iximab 2, golimumab 1, and tocilizumab 2), whereas 3 patients switched from infl iximab to golimumab, tocilizumab, or abatacept. The study was approved by the ethics committee of our institute, and all patients gave their written informed consent. Multimodality imaging assessments, including X-ray, 18 F-FDG and 18 F-NaF PET/CT, of bilateral hands and wrists were completed within 4 days prior to starting or switching bDMARDs. In addition, MRI was performed in all patients except one, who had a permanent pacemaker. For 6 months after the fi rst imaging assessment, we monitored swollen joint counts, tender joint counts, patient ' s and physician ' s global assessments, patient ' s pain assessments, and modifi ed health assessment questionnaires (MHAQs) [27]. Using the erythrocyte sedimentation rate (ESR), we calculated the disease activity score in 28 joints (DAS28) according to the established formula [28]. To assess bone turnover, we measured serum osteocalcin, serum bone alkaline phosphatase, and total urinary deoxypyridinoline levels at baseline.

Hand X-ray
Radiographs of joints in the hands and wrists were assessed at baseline and 6 months according to the Genant-modifi ed Sharp scoring system [29] by three independent readers blinded to the treatment assignment, clinical fi ndings, and chronologic order of radiographs. The inter-reader agreement was acceptable ( κ ϭ 0.78). Total radiographic score was composed of erosion score plus joint space narrowing (JSN) score. Each site was evaluated in 0.5-unit increments. Erosion scores of 0 -3.5 are assigned to 14 sites in each hand and wrist. JSN scores of 0 -4 are assigned to 13 sites in each hand.

PET scanning
18 F-FDG and 18 F-NaF PET/CT scans were performed using a SET 2400 W (Shimadzu, Kyoto, Japan). Blood glucose levels were lower than 150 mg/dL after fasting more than 6 h when 2.5 MBq/ kg of 18 F-FDG was injected into the patients. After 60 min of an uptake phase, data of PET in the bilateral hands were acquired for 2 min in prone with arm up position. More than 24 h later 18 F-NaF PET scan was conducted for 2 min scanning of the hands after injection with 185 MBq of the tracer followed by 40 min of uptake. A multislice helical non-contrast CT scan was obtained prior to each PET scanning and used for attenuation correction and anatomic information of the PET images. The spatial resolution of the PET was 2.5 mm pixel size in this study. Experienced radiologists determined hypermetabolic areas in bilateral metacarpophalangeal (MCP), proximal interphalangeal (PIP), and wrist (radiocarpal, ulnocarpal, and intercarpal) joints. The maximum standardized uptake value (SUVmax) in individual joints was determined according to the following equation: SUVmax ϭ maximal count ϫ calibration factor (kBq/mL)/ injected activity (MBq)/body weight (kg). SUVmax in the wrist joint was determined by the highest SUVmax among those of the radiocarpal, ulnocarpal, and intercarpal joint regions. The sum of the SUVmax for all 26 joint regions was calculated.

MRI
As an option, we performed MRI mainly as a reference to other imaging modalities. Plain MRIs of the wrist and fi nger joints were acquired using a 1.5-Tesla scanner (Gyroscan Intera Master, Philips Medical Systems, Eindhoven, Netherlands), according to the institute ' s routine procedure at study of the entry. The imaging protocol comprised a coronal short T1 inversion recovery sequence (repetition time, 4,000 -4,200 ms; echo time, 80 ms), followed by coronal and axial T1-weighted spin echo images (repetition time, 500 -600 ms; echo time, 11 ms). The slice thickness was 3 mm.

Statistical analysis
Correlations were investigated using Spearman ' s rank correlation. Chi-square and unpaired t-tests were used to compare radiopharmaceutical uptake in the joints with and without progressive erosions. P values less than 0.05 were considered statistically signifi cant. A stepwise multivariate regression analysis was conducted to test the independent determinants of the estimated yearly radiographic progression. The independent variables included tender joint count, swollen joint count, patient ' s global assessment, physician ' s global assessment, DAS28-ESR, MHAQ, C-reactive protein (CRP), ESR, matrix metalloproteinase-3 (MMP-3), osteocalcin, bone alkaline phosphatase, urinary deoxypyridinoline, sum of the SUVmax of 18 F-FDG, and sum of the SUVmax of 18 F-NaF. All statistical analyses were performed in SPSS version 11.0 (IBM Japan, Tokyo, Japan). Table 1 summarizes the baseline clinical profi les of the patients in this study. All patients were positive for ACPA. The disease duration was from 3 months to 22 years. No patients had achieved clinical remission (based on DAS28) at entry, despite previous treatment with low-dose prednisolone (4 -15 mg/day) in 6 patients, methotrexate (6 -16 mg/week) in 9 patients, and infl iximab in

F-FDG and F-NaF uptake in individual joints
First, we analyzed the 18 F-NaF and 18 F-FDG PET/CT images at entry. In a 74-year-old male with a 13-year history of RA ( Figure 1A), both tracers accumulated in the 2nd MCP joint, which had swelling and tenderness and bone erosion in the metacarpal bone head on the plain radiograph. 18 F-NaF signals were located on the bone cortex, whereas 18 F-FDG signals were mainly in the joint space adjacent to the bone. Thus, upregulated bone turnover is likely associated with active infl ammation. In contrast, 18 F-NaF accumulated not only in joints with erosion but also in the 3rd distal interphalangeal (DIP) joint. According to radiography, osteophyte as well as JSN and subchondral sclerosis was observed in this joint, which was considered to be an OA change ( Figure 1B).
We compared the uptake of both tracers in 26 joint regions, including all PIP and MCP joints and radial, medial, and ulnar regions of the wrists. 18 F-FDG and 18 F-NaF accumulated in 73%  13-year history of RA. Bone erosion was found in the 2nd metacarpal bone head in plain X-ray. 18 F-NaF signals were located on the bone cortex, while 18 F-FDG signals were observed in the joint space. (B) A 68-year-old female with a 3-month history of RA. 18 F-NaF accumulated not only in joints with erosion but also in the 3rd DIP joint. According to radiography, osteophyte as well as JSN and subchondral sclerosis was observed in this joint, which was considered to be an OA change.
radiographic scores signifi cantly correlated with the sum of the SUVmax of 18 F-NaF ( r ϭ 0.69, p ϭ 0.014) but not 18 F-FDG ( r ϭ 0.05, p ϭ 0.879) (Supplementary Table 2 to be found online at http://informahealthcare.com/doi/abs/10.3109/14397595.2015. 1069458), and was elevated by more than 0.5 points in 10 patients, unchanged in another, and reduced in one. Sum of the SUVmax of 18 F-NaF was signifi cantly correlated with the progression of bone erosion ( r ϭ 0.58, p ϭ 0.045) but not with JSN ( r ϭ 0.47, p ϭ 0.12) in two components of the total radiographic score. A stepwise multivariate regression analysis was carried out to test the independent determinants of radiographic progression. We found that the SUVmax of 18 F-NaF ( β ϭ 0.66, p ϭ 0.028) were the only variables independently associated with radiographic progression among clinical backgrounds, physical fi ndings, disease activity, laboratory data including bone metabolism markers, and PET fi ndings (Table 2).
We then analyzed 18 F-NaF and 18 F-FDG PET in individual joints, which were divided into subgroups according to interval change for the 6-month observation. Erosive joints ( n ϭ 44) at the baseline were divided into persistent ( n ϭ 12), progressive ( n ϭ 22), and repaired ( n ϭ 10) groups; other groups were non-erosive joints ( n ϭ 268) including unchanged joints ( n ϭ 246), and newly developed erosion ( n ϭ 22) ( Figure 3A). SUVmax for 18 F-NaF was the highest in the progressive erosion group, followed by the persistent erosion group; SUVmax for 18 F-FDG was also signifi cantly higher in the progressive erosion group than in the other groups except the repaired group ( Figure 3B). Accumulation of both tracers in a particular joint was divided into 4 patterns: 18 F-NaF ϩ / 18 F-FDG ϩ , 18 F-NaF ϩ / 18 F-FDG-, 18 F-NaF-/ 18 F-FDG ϩ , and 18 F-NaF-/ 18 F-FDG-. The 18 F-NaF ϩ / 18 F-FDG ϩ pattern was signifi cantly more frequent in the progressive erosion joints than in any other subgroup; however, 18 F-NaF-/ 18 F-FDG-was the most dominant pattern in the non-erosive joints ( Figure 3A). Therefore, bone lesion progression was likely associated with the uptake of both tracers in the joints.

Comparison with interval changes of MRI
We conducted follow-up 18 F-FDG and 18 F-NaF PET/CT and MRI at 6 or 12 months in 2 patients ( Figure 4). In Case 3 ( Figure 4A), golimumab was initiated at study entry because high disease activity (DAS28 5.69) had been persistent. The patient showed a moderate response to golimumab (DAS28 4.72), but clinical effi cacy was not apparent in the left wrist, in which tenderness and swelling remained at 6 months. The baseline 18 F-FDG accumulated mainly in the ulnocapitate and radioscaphoid joint spaces; however, diff use 18 F-NaF uptake was found in the carpal bones but not the radius or ulna. These signals remained in the follow-up PET examinations at 6 months, and the MRI detected a newly developed erosion in the lunate bone.
In contrast, Case 4 ( Figure 4B) achieved clinical remission (DAS28 2.91 at baseline to 1.42 at 6 months) with etanercept and maintained remission at 12 months. The right wrist joints also showed clinical improvement. 18 F-NaF uptake was diff use throughout the carpal bones of the left wrist joint, whereas weak 18 F-FDG accumulation was present in the joint spaces at baseline. At the 12-month follow-up examination, 18 F-FDG signals were almost undetectable; however, signifi cant 18 F-NaF signals remained in two regions: the carpal articular surface of the radius and the joint surface of the capitate. In accordance with the remaining 18 F-NaF signals, erosions were repaired. Therefore, 18 F-NaF accumulation without 18 F-FDG signals can be detected in bone undergoing repair.

Discussion
In the present study, 18 F-FDG and 18 F-NaF PET/CT illustrated the aff ected joints of RA patients; overall accumulation of 18 F-NaF correlated with clinical assessment and physical disability. According to comparative analysis with radiographic images, and 82% of swollen joints and 16% and 30% of non-swollen joints, respectively. The SUVmax values of 18 F-FDG and 18 F-NaF were signifi cantly correlated in individual joints ( Figure 2A); however, there was a discrepancy in accumulation in some joints.
We separately analyzed radiographic OA change-positive ( n ϭ 47) and -negative ( n ϭ 265) joints in RA-prevalent joint regions ( Figure  2B). There was a signifi cant correlation between the SUVmax of the two tracers, irrespective of the presence or absence of OA change.
18 F-FDG and 18 F-NaF PET uptake and clinical assessments at baseline 18 F-FDG and 18 F-NaF peak uptake parameters, which were calculated by summation of the SUVmax in 26 joint regions, were 20.3 Ϯ 15.4 and 62.1 Ϯ 35.3, respectively. The sum of the SUVmax of 18 F-NaF, but not 18 F-FDG, correlated with clinical disease activity, according to DAS28 and physical function assessed by MHAQ at baseline (Supplementary Table 1 to be found online at http://informahealthcare.com/doi/abs/10. 3109/14397595.2015. 1069458). In addition, the sum of the SUVmax of 18 F-NaF correlated with MMP-3, but neither score correlated with other laboratory parameters.
All patients underwent the follow-up X-ray imaging of the hands at 6 months. The estimated yearly progression of total 18 F-NaF accumulation was associated with the presence of erosive lesions and ongoing bone damage; 18 F-FDG uptake also refl ected this damage. In contrast, 18 F-NaF and 18 F-FDG signals were not detected in most of the intact joints. The concordant distribution of both tracers in the same joints suggests coupling of infl ammation with upregulated bone turnover in RA-aff ected joint lesions. Although the patterns of 18 F-NaF and 18 F-FDG accumulation did not exactly overlap, the infl ammation and bone turnover reveal both lead to bone destruction.
A number of 18 F-FDG PET studies have shown that 18 F-FDG is incorporated into the infl ammatory cells of soft tissues, including macrophages, capillaries, and fi broblasts; osteoclasts also take up the tracer in bone [30], and that the accumulation in RA-aff ected joint lesions correlates with disease activity [14 -18]. 18 F-NaF is directly incorporated into the areas of bone with upregulated turnover; however, blood fl ow is a rate-limiting step of the tracer ' s uptake [24], indicating that this imaging technique has potential to refl ect two main pathologic components: hypervascularity and bone destruction in RA. Furthermore, in experimental animal arthritis, 18 F-NaF PET shows a correlation between the tracer accumulation and progression of bone destruction [25]. Irmler IM et al. reported the relationship of 18 F-FDG and 18 F-NaF uptakes and joint destruction [25,31]. 18 F-FDG uptakes, which correlated with arthritis score, start to increase at Day 9 and reach the peak at Day 11 followed by decline. On the other hand, 18 F-NaF signals, which correlated with bone destruction, were elevated and sustained after Day 11. The maximum incorporation was noted from Day18 to 25, when arthritis has already subsided. Thus, the data suggest that infl ammation is followed by bone destruction.  Figure 3. 18 F-FDG and 18 F-NaF PET/CT fi ndings and radiological changes of bone lesions. 18 F-FDG and 18 F-NaF PET/CT fi ndings were compared among 5 groups, which were divided according to radiological fi ndings at the baseline and interval changes for 6 months.
(A) Signal intensities of 18 F-FDG and 18 F-NaF were signifi cantly higher in erosive joints than non-erosive ones, and were the highest in joints having erosion with further progression. (B) In analysis of accumulation patterns of 18 F-FDG and 18 F-NaF, simultaneous accumulation of the both tracers was signifi cantly more frequent in joints having erosion with further progression than any other groups. No 18 F-NaF signal was detected until clinical fi ndings of the joint lesions appeared; therefore, autoimmune infl ammation is followed by bone lesions [25]. Bone scintigraphy using 99mTc-labeled bisphosphonate has been used as a nuclear imaging technique to detect joint infl ammation in RA until 18 F-NaF PET is available [32]. Accumulation of 99mTc bisphosphonate, which is found in periarticular bone lesions with osteolytic lesion and hypervascularity, predicts later erosion [33]. Specifi city to detect bone lesions is greatly improved by introduction of SPECT/CT, but further signifi cantly improved with use of 18 F-NaF PET/CT [23,34].
In concordance with previous reports, 18 F-NaF was preferentially incorporated into erosive lesions in our study; the highest signals were associated with progressive erosion [24]. Furthermore, the overall accumulation correlated with progression of erosion (Supplementary Figure 1 to be found online at http:// informahealthcare.com/doi/abs/10.3109/14397595.2015.1069458. Numerous studies show that total US-PD signals are associated with bone destruction, especially in cases of longer lasting, more active infl ammation [35 -37]. However, 18 F-NaF PET more directly detects ongoing bone destruction in RA-aff ected joints compared with other imaging modalities. We compared fi ndings in individual joints of hands between 18 F-NaF and 18 F-FDG PET/CT scans. There is a close correlation between 18 F-FDG and 18 F-NaF accumulations in the joints of RA patients irrespective of OA changes. However, typical OA lesions ( Figure 1B, in DIP joints) incorporated 18 F-NaF but not 18 F-FDG, though DIP joints were not systematically assessed in this study. Moreover, according to previous reports, some OA lesions take up 18 F-FDG [34,38]; therefore, secondary synovitis is likely accompanied by OA. Alternatively, RA synovitis may be complicated osteoplastic changes in our patients, because the observation was restricted to RA-prevalent joints.
Combining PET with CT scanning is helpful for identifying anatomical localization of PET signals. The anatomic distributions of the markers in some joints diff ered between 18 F-NaF and 18 F-FDG ( Figure 1A). Specifi cally, 18 F-NaF accumulated in erosive lesions of the bone cortex; 18 F-FDG signal was located in the surrounding soft tissues. These diff erential distributions are reasonable considering the pharmacologic features of the tracers. However, the spatial resolution of PET/CT was insuffi cient to identify detailed localization in many cases, especially when the signals are too intense [39]. The issue is one of the major limitations of PET/CT in small joint analysis. Recent pilot studies have shown that spatial resolution of PET image in hand small joints is improved by use of dedicated PET imaging devices [34], or high-resolution PET with MRI scanner [40].
The present study has several limitations due to the study design and nature of 18 F-NaF as a tracer. We enrolled RA patients who required the fi rst or second bDMARD with a wide range of disease duration; however, the sample size was too small to discuss the eff ects of various background factors (such as disease stage and therapies) on fi ndings of PET fi ndings.
18 F-NaF is incorporated into not only RA-associated bone destructive lesions but also osteoplastic lesions caused by other disorders, such as OA. Moreover, strong 18 F-NaF signals were found in areas of bone erosion repair in a patient with a favorable clinical response to etanercept ( Figure 4B, Case 4). Case 4 showed that remaining 18 F-NaF signals refl ected undergoing repair. This is compatible with the fi nding that 18 F-NaF signals last longer than the 18 F-FDG signals after subsiding arthritis in experimental animal models. Erosive progression occurred simultaneously with repair in other joints in the same patients, as we previously reported [41]. Therefore, it is diffi cult to determine the pathology on the basis of the fi nding of 18 F-NaF PET alone. Rather, comprehensive assessment together with other modalities is essential to characterize the pathology of 18 F-NaF-positive sites. Another concern is radiation exposure in clinical application; repeated examinations are not recommended because computed tomography dose index volume is estimated as 9.9mGy per test. The radiation dose for a whole-body study is 25 mGy.
In summary, co-existing PET signals of 18 F-NaF and 18 F-FDG in the aff ected joints suggest that infl ammation is coupled with upregulated bone turnover, leading to joint destruction. In particular, joints with strong signals of both tracers are likely to have progressive bone destruction in the near future.