Dissociation between 2-[18F] uoro-2-deoxy-D- glucose Positron Emission Computer Tomography, Ultrasound and Clinical Assessments in Patients with non-severe Rheumatoid Arthritis including Remission

Charline Rinkin (  c.rinkin@chuliege.be ) Centre Hospitalier Universitaire De Liege https://orcid.org/0000-0003-2243-6799 Pacôme Fosse Centre Hospitalier Universitaire d'Angers Olivier Malaise Centre Hospitalier Universitaire De Liege Nathalie Chapelier Centre Hospitalier Universitaire De Liege Jil Horrion Centre Hospitalier Universitaire De Liege Laurence Seidel Centre Hospitalier Universitaire De Liege Albert Adelin Centre Hospitalier Universitaire De Liege Roland Hustinx Centre Hospitalier Universitaire De Liege Michel G. Malaise Centre Hospitalier Universitaire De Liege


Background
Since the use of biologic agents in the therapeutic armamentarium of rheumatoid arthritis (RA), low disease activity (LDA) and remission are reachable targets with better outcomes including less radiographic progression (1). In clinical trials, remission is often de ned as a disease activity 28-joint score (DAS 28 ) <2. 6. Although relatively lenient, this target is achieved in only 30-40 % of the patients (2)(3).
Within this group, the disease remains active in a signi cant proportion of patients as observed in the DREAM registry where 31.1% of patients had a swollen joint count ≥2 (3) and experienced joint damage progression (4). Further, imaging studies have shown at least one synovitis in 33-73 % of patients in remission by ultrasound (US) and in up to 96% of them by magnetic resonance imaging (MRI) (5).

F] uoro-2-deoxy-D-glucose Positron Emission
Computer Tomography ([ 18 F]FDG PET/CT) is capable to visualize and quantify in ammation in RA synovitis. The number of PET-positive joints analyzed on the 28 joints of the DAS and the cumulative standard uptake value (CSUV) of these PET-positive joints were highly correlated with the clinical status (number of swollen joints, number of tender joints) and with erythrocyte sedimentation rate (ESR) and Creactive protein (CRP) parameters, with the DAS 28 (22,24) but also with US data such as the number of US-positive joints and the cumulative synovial thickness (22). Roivanen et al (21) reported that up to 90% of the joints rated positive by clinical evaluation (swollen and tender) were also positive on [ 18 F]FDG PET whereas a proportion of 75% was quoted by Elzinga (19). In these studies however, correlation between PET and the DAS 28 scores was exclusively obtained in RA patients with severe disease activity (20,22) or in a large majority of them (21,24). The primary objective of the present study was to compare PET/CT parameters among RA patients under remission, under low/moderate or under severe activity as de ned by two classical composite indices, the DAS 28 -CRP score and the Clinical Disease Activity Index (CDAI).

Study design and patients
The cross-sectional study, approved by the ethical committee of our hospital (B70720108722), included 63 patients ful lling the ACR/EULAR 2010 criteria for RA (25), from July 2010 to April 2012. Written informed consent was obtained for each patient. All assessments were done on the same day by the same independent experienced investigator unaware of the other results: the clinical evaluation rst in the morning with a blood sample followed by the US and terminated by the PET/CT evaluation. X-rays were available as routine controls made at maximum 6 weeks around the study day. The Patient (PtGA) and the Physician Global Assessment (PGA) were determined using a Visual Analogue Scale (VAS) (0-100 mm) as well as the Health Assessment Questionnaire (HAQ) (26). Disease activity was evaluated using the DAS 28 -CRP (lacking PGA) (27) and the CDAI (with PGA and without CRP) (28). Each patient was categorized as in remission (DAS 28 -CRP<2.6, CDAI<2.8), in low to moderate disease activity (2.6-DAS 28 -CRP≤5.1, 2.8-CDAI≤22), or in severe disease activity (DAS 28 -CRP>5.1, CDAI>22) (29). The number of joints solely tender (T), solely swollen (S), "tender or swollen" (T/S) and "tender and swollen" (T&S) was recorded.
[ 18 F]FDG PET/CT imaging The PET/CT studies were performed using a Gemini BigBore scanner (Philips Medical Systems, Cleveland, OH, USA). Patients fasted for 4 hours and were injected with [ 18 F]FDG through an indwelling catheter placed in the median cubital vein and ushed with 5cc of saline solution afterwards (4 MBq/kg body weight with a maximum of 370 MBq). Blood glucose level was lower than 140 mg/dl. The uptake time was 60 minutes and the image acquisition sequence as follows: rst a scoutview CT, followed by the PET emission study that included the knees, hands, wrists, elbows and shoulders, with 2 minutes per bed position for a total scanning time ranging from 14 to 18 minutes. Finally, a low-dose CT (5-mm slice thickness, tube voltage 120 Kv, tube current-time product 80 mAs) was performed over these joints. The hands and wrists were positioned and xated on a dedicated Plexiglas device in order to avoid movements between the PET and the CT acquisitions. PET Images were reconstructed using an iterative list mode time-of-ight algorithm and corrections for attenuation, dead-time, random and scatter events were applied. The images were rst analyzed visually and joints were considered as positive for synovitis when the [ 18 F]FDG uptake was increased compared to the background in areas corresponded to joint synovium on CT, i.e. either when thickened synovium was recognized on CT or in locations corresponding anatomically to synovium, excluding uptake in other structures such as muscle and tendons. The [ 18 F]FDG uptake was then quanti ed using the maximum standardized uptake value (SUVmax). In PETpositive joints according to the visual analysis, the SUVmax was obtained by drawing a region of interest (ROI) over the most active synovial area identi ed. When no synovitis was identi ed, ROIs were placed in the corresponding areas on the CT: at the dorsal surface of the radius (on top of the lunate) for the wrists, over the lateral recess at the level of the midpatella for the knees and for the small joints as metacarpophalangeal (MCP) or proximal interphalangeal (PIP) joints, ROIs were drawn around the appropriate joint. A global metabolic assessment was obtained through the number of PET-positive joints (visual evaluation) and the sum of all SUVs (cumulative SUV, CSUV).
Ultrasound and X-ray examination US assessment was performed using a B-mode multi frequency 10-14.0 MHz transducer (Logiq 9) (GE Healthcare, Milwaukee, WI, USA). US positioning for the wrists, the MCP and the PIP joints and for the knees have been described elsewhere [22]. Proximal and distal radiocapitellar recesses were studied for the elbows and the glenohumeral joint, posterior side for shoulders. Synovial measurements were carried out systematically perpendicular to the great axis and at the point of greatest thickness. A cut-off for US positivity was de ned as a synovitis of at least 1 mm thick (3 mm for the shoulders) according to US determinations in healthy controls, described elsewhere (22). In joints where 2 (wrists) or 3 (elbows, knees) scannings were performed, the joint was considered positive if at least one measurement was positive. The cumulative synovial thickness (CST), i.e the sum of thicknesses of all US-positive joints, is the addition of all (single or multiple) synovial measurements performed. X-ray were obtained for peripheral joints (PIPs, MCPs and wrists)

Statistical analysis
Results were generally expressed as mean ± standard deviation (SD). Correlation coe cients were calculated to measure the association between PET/CT and clinical or US parameters. The Spearman correlation was used for skewed distributions. Concordance between methods was quanti ed by the intraclass coe cient (ICC). Ordinal logistic regression was used to assess the relationship between disease activity categories based on DAS 28 -CRP or CDAI (remission, low/moderate and severe disease activity), and PET/CT number of positive joints and CSUV. A test was performed whether all three diseases severity categories were distinguishable. If this was not the case, a classical logistic regression analysis was applied and optimal Youden cut-off values were determined from the Receiver Operating Characteristic (ROC) curve method. Results were considered signi cant at the 5% level (p<0.05). All statistical analyses were performed with SAS (version 9.4).

Relationship between PET/CT and clinical measurements
Joint positivity on PET/CT was compared to clinical evaluation (tender, swollen, "tender or swollen", "tender and swollen") for each joint. Diagnostic e cacy of PET/CT (sensibility, speci city, positive and negative predictive values) is presented in Table 2. PET/CT sensitivity was low with respect to clinical measurements: only 58.9% of the joints that were both tender and swollen, and only 35.8% of the joints that were tender or swollen, were positive on PET/CT. Speci city was higher: 82.6% of the joints that were not tender or not swollen, and 85.4% of the joints that were neither tender nor swollen, were PET-negative (Table 2). In accordance, the positive predictive value of PET-CT was low, while the negative predictive value was as high as 96.2% in tender and swollen joints. In other terms, when PET-CT was negative, the probability that the articulation was not tender and/or not swollen was high, while when PET-CT was positive, the probability that this articulation was tender and/or swollen was of poor value.

Relationship between PET/CT parameters and disease activity threshold
The number of PET/CT-positive joints and CSUV were analyzed according to disease activity categories (based on DAS 28 -CRP or CDAI) and illustrated in Figure 1 and Supplementary Table 1 (e.g. there were respectively 3.6 ± 5.4 PET/CT-positive joint for DAS28-CRP remission, 4.7 ± 6.7 for low/moderate activity and 13.6 ± 11.2 for severe disease activity). An ordinal logistic regression evidenced a signi cant relationship between the mean number of PET/CT-positive joints or the CSUV and clinical disease activity (Supplementary Table 1). However, "remission" and "low/moderate" disease activity categories could not be dissociated by PET/CT (the cutoff value between the two groups was negative, so that patients in remission should have a negative number of PET/CT positive joints which is impossible). Moreover, 27.3% of patients without any metabolic activity were observed in both DAS 28 -CRP and CDAI remission, while 25.8% and 27% in DAS 28 -CRP and CDAI low/moderate activity sub-groups (p=0.99 for both DAS 28 -CRP and CDAI subgroups) indicating that PET/CT was unable to discern remission and low-moderate activity (Supplementary Table 2).
Thus, the two categories remission and low/moderate were merged and a classical logistic regression analysis was performed between patients with severe and non-severe (including remission and low/moderate activity) disease activity (Table 3): highly signi cant differences were found in the number of PET/CT-positive joints and in CSUV. The optimal threshold for identifying RA patient with a clinically and biologically severe disease was at least 8 PET/CT-positive joints and a CSUV ≥ 17.8 for the DAS 28 -CRP and 6.8 and 15.0 for the CDAI, respectively. Disease activity thresholds were also studied by dividing RA patients into remission and non-remission categories (including low/moderate and severe disease), but no signi cant difference was observed in terms of number of PET/CT-positive joints and CSUV (data not shown).
Supplementary Table 2  Overall there was a signi cant correlation between the metabolic measurements (number of positive joint and CSUV) and the clinical assessments (DAS 28 -CRP and CDAI) ( Table 4). However, when classifying patients in remission, low/moderate or severe category, this signi cant correlation between PET and clinical assessment was observed only in the subjects with severe activity according to CDAI, and not in RA patients in low/moderate disease activity or remission.

Relationship between X-rays and PET/CT
The 63 peripheral joints (PIPs, MCPs and wrist) that were PET-positive but US-negative in the 16 patients in clinical remission (DAS28-CRP<2.6) were characterized with X-ray. Features of RA, i.e. symmetrical joint narrowing, bone erosion or demineralization, and of OA, i.e. asymmetrical joint narrowing, subchondral sclerosis, or osteophytes were recorded. Results were consistent with RA in 24/63 joints (38,1%) and with OA in 8/63 joints (12.7%) with OA signs. In 31/63 joints (49.2%) X-rays were normal. In particular, RA / OA / normal feature was described in 7 / 0 / 24 of the 31 PIPs, 12 / 6 / 5 of the 23 MCPs and 5 / 2 / 2 for the 9 wrists. Considering the corresponding clinical status, none of these joints were tender or swollen.

Discussion
In line with previous work (22), we con rmed that the number of PET/CT-positive joints and the CSUV correlate signi cantly with the number of US-positive joints, synovial thickness and disease activity based on either on DAS 28 -CRP or CDAI. In addition, PET/CT was quite effective at distinguishing patients with a severely active disease from the others, as a cutoff of 8 for the number of PET-positive joints and 17.8 for the CSUV yielded an area under the curve (AUC) of 0.77 (considering DAS 28 -CRP as the clinical gold standard). Although the number of hypermetabolic joints and the cumulative SUV tended to be higher with increased clinical severity of the disease, one notable exception should be mentioned. There was no signi cant difference in the number of PET-positive joints and their CSUV between patients in clinical remission and those with a low/moderate disease activity. In both groups, only 25-27% of the patients presented with perfectly negative PET/CT ndings. Clearly, PET/CT results and clinical assessment diverge in non-severe RA including remission, in agreement with previous observations made with US and MRI (3)(4)(5)(6)(7)(8)(9).
Comparing the PET and the US ndings, there were twice as many PET/CT positive joints as US-positive joints. Furthermore, there was also clear evidence that PET/CT and US analysis of joints did not concur in remission. For example, out of the 22 patients in remission by DAS 28 -CRP, 12 were positive with both PET/CT and US, but at the joint level, only 10 of the 75 PET/CT-positive joints (5 wrists, 4 MCPs and one shoulder) were also US-positive. This divergence was also observed in patients in Boolean remission. In other words, PET/CT was positive in a signi cant number of patients with no or low/moderate disease activity according to the current clinical scales, and was also positive in a signi cant number of joints that were not considered as in amed according to clinical and US parameters. Two interpretations are possible for this observation. The rst one would be a higher sensitivity of the metabolic measurements for identifying subclinical joint in ammation. Indeed, in in ammatory diseases, PET/CT incidental ndings due to [ 18 F]FDG accumulation is consistently associated with an enhanced glycolytic metabolism in in ammatory cellular in ltrates including activated macrophages, neutrophils and lymphocytes (30). We may therefore consider that hypermetabolic joints with normal US appearance are joints with an in ammatory component without proliferating synovitis or with a synovitis < 1 mm thickness, the cut-off retained. In a previous series of RA patients with severe disease activity, only 50% of the PIPs and 62% of the MCPs both tender and swollen were US-positive using the same cut-off (data not shown) (22). PET/CT analysis might therefore exhibit greater sensitivity than US. It is noteworthy that in the current series, the PET/CT-positive but US-negative joints within these 16 patients were mostly PIPs (31 joints in 7 patients) and MCPs (23 joints in 7 patients), joints typically involved in RA. X-ray analysis supports this hypothesis as 38% of the joints had signs of RA and 49% were normal. An alternative explanation would be to consider those joints and patients as false positive results of the PET/CT. It is indeed possible that the joints actually suffer from secondary (MCPs) or primary (PIPs) osteoarthritis. However, only 8/62 (13%) joints, 6 MCPs in 2 patients and 2 wrists in 2 patients had signs of OA. The radiological analysis is thus in favor of the rst hypothesis but a longitudinal follow-up of the patients would be needed to provide de nitive evidence. As a limitation, X-ray were only available for peripheral joints ((PIPs, MCPs and wrist) with systematic X-ray realisation, but not for larger joints (knee, elbow, shoulder). [ 18 F]FDG PET/CT demonstrated a high speci city and negative predictive value compared to individual clinical evaluation of the joints. Furthermore, PET/CT is effective at differentiating "severe" from "nonsevere" patients, although clinical remission was not associated with metabolic remission. Such issues are of high clinical relevance as PET/CT could possibly identify subclinical and infra-radiological in ammation worthy of treatment in order to prevent further irrevocable damages to the joints. Further studies are needed to ascertain whether this represents a clinically relevant activity of the disease or secondary degenerative changes. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Conclusion
The study was approved by the ethical committee of the university hospital of Liège (B70720108722).
Informed consent was obtained from all individual participants included in the study.

Consent for publication:
Not applicable Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests:
The authors declare that they have no competing interests Funding: This work was supported by academic funding from "Fonds Léon Fredericq", University of Liège, Belgium. "Fonds Léon Fredericq" had no implication in the acquisition or analyzing of the data.

Authors'contributions:
CR performed the study, the US data acquisition, analyzed and interpreted the results wrote the manuscript and revised the manuscript critically. PF performed the PET-CT data acquisition, analyzed and interpreted the results and revised the manuscript critically. OM analyzed and interpreted the results, wrote the manuscript and revised the manuscript critically. NC performed the clinical evaluation and revised the manuscript critically. JH performed the X-ray data acquisition and revised the manuscript critically. LS and AA analyzed and interpreted the results, realized all the statistics analyses, and revised the manuscript critically. RH and MM designed the study, analyzed and interpreted the results, wrote the manuscript and revised the manuscript critically. All the authors approved the nal manuscript for submission and publication. All co-authors take full responsibility for the integrity of the study and all parts of the nal manuscript.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.