Synovial fibroblast activation occurs before the onset of rheumatoid arthritis and influences the risk of developing disease Authors: Armin Atzinger, Koray Tascilar, Arnd Kleyer, Axel J Hueber, Christian Schmidkonz, Torsten Kuwert, Michael Beck, Filippo Fagni, Giulia Corte, Andreas Ramming, Georg Schett, Sara Bayat Published: 2025 ABSTRACT AbstractObjectivesActivated synovial fibroblasts play a key role in rheumatoid arthritis (RA). Positron emission tomography-CT (PET-CT) using 68gallium-labelled fibroblast activation protein inhibitor (FAPI) allows the detection of FAP-expressing activated fibroblasts in humans in vivo. Herein, this study aimed to investigate whether fibroblast activation already occurs in the pre-clinical phase of RA and whether it is associated with the development of clinical disease.Methods68Ga-FAPI-46 PET-CT was performed in individuals positive for anti-citrullinated protein antibodies (ACPAs) with clinically suspect arthralgia without present or past signs of joint swelling or present or past anti-rheumatic therapy. All participants underwent structured clinical and laboratory evaluations and were followed for the development of RA. Synovial FAPI uptake was quantified and related to demographic characteristics, joint tenderness, ACPA levels and RA development.FindingsEighteen RA-at-risk individuals (male/female: 7/11; mean age, 47 (SD 14) years) underwent 68Ga-FAPI-46 PET-CT and were followed for a median of 42 weeks. Five participants (28%) developed RA with a median of 12 weeks (IQR, 11–20), whereas 13 remained disease-free. Total lesion FAPI-uptake (TL-FAPI) was a significant predictor of RA development (HR 4.58, 95% CI 1.38 to 15.22, p=0.013). Increasing TL-FAPI at the joint level was also associated with tenderness (isk ratio 1.14, 95% CI 1.08 to 1.22, p=0.001).ConclusionsThe findings in this small RA-at-risk cohort suggest that synovial fibroblast activation occurs early in the disease process of RA and is associated with an increased risk for clinical RA onset. These results warrant further research on FAPI-PET-CT to improve risk assessment for clinical RA onset. WHAT IS ALREADY KNOWN ON THIS TOPIC Fibroblast activation plays a key role in the development of synovitis in patients with rheumatoid arthritis (RA). Fibroblast activation plays a key role in the development of synovitis in patients with rheumatoid arthritis (RA). Activated fibroblasts can be visualised in vivo by positron emission tomography-CT using 68gallium-labelled fibroblast activation protein inhibitor (8Ga-FAPI). Activated fibroblasts can be visualised in vivo by positron emission tomography-CT using 68gallium-labelled fibroblast activation protein inhibitor (8Ga-FAPI). RA is characterised by a pre-disease state in which individuals develop anti-citrullinated protein antibodies (ACPAs) and clinically suspect arthralgia. RA is characterised by a pre-disease state in which individuals develop anti-citrullinated protein antibodies (ACPAs) and clinically suspect arthralgia. WHAT THIS STUDY ADDS Synovial 68Ga-FAPI-46 uptake is a frequent finding in patients at risk for RA development, suggesting that synovial fibroblast activation occurs before clinical disease onset. Synovial 68Ga-FAPI-46 uptake is a frequent finding in patients at risk for RA development, suggesting that synovial fibroblast activation occurs before clinical disease onset. This study points out that on a per patient basis, increased FAPI uptake in the joints is associated with higher risk of developing RA. This study points out that on a per patient basis, increased FAPI uptake in the joints is associated with higher risk of developing RA. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study supports the critical role of synovial fibroblast activation in RA development, showing that synovial fibroblast activation is part of the pre-disease state and influences RA development. Further research on fibroblast imaging to improve risk assessment of RA onset is warranted. This study supports the critical role of synovial fibroblast activation in RA development, showing that synovial fibroblast activation is part of the pre-disease state and influences RA development. Further research on fibroblast imaging to improve risk assessment of RA onset is warranted. Introduction Rheumatoid arthritis (RA) is a chronic inflammatory joint disease, characterised by synovitis and bone and cartilage damage.1 2 Early identification of patients with RA is crucial to initiate effective treatment. Synovial inflammation in RA is characterised by fibroblast activation (‘synovial hyperplasia’) and the infiltration of innate and adaptive immune cells. The current understanding of RA development is based on a stepwise process that begins with autoimmunity against immunoglobulins and post-translationally modified proteins, evidenced by the emergence of rheumatoid factors (RF) and anti-citrullinated protein antibodies (ACPAs) that precede the onset of clinical disease by years.3 Before RA onset, joint pain increases, which has been described as ‘clinically suspect arthralgia’.4 Individuals showing ACPA and clinically suspect arthralgia are at high risk for RA development and usually show subclinical signs of inflammation in ultrasound or MRI.57 Clinical studies have shown that intervention with methotrexate or abatacept in individuals at high risk for developing RA improves early symptoms, with abatacept also inhibiting the progression to RA.810 Fibroblasts are a central component of RA, influencing inflammation through the production of cytokines, chemokines and growth factors that allow the attraction and activation of immune cells.11 12 Furthermore, synovial fibroblasts bear a local inflammation memory regulating the anatomical distribution of inflammation and above that intimately contribute to bone and cartilage damage in RA.13 Assessing fibroblast activation in humans in vivo has been challenging until recently; however, new theranostic compounds that specifically bind to activated fibroblasts have allowed the use of advanced radiopharmaceuticals for this purpose. This development made use of the observation that activated fibroblasts, unlike their quiescent counterparts, express markers such as fibroblast activation protein (FAP), a transmembrane serine protease with dipeptidyl peptidase and endopeptidase activity.14 This technique has allowed us to define fibroblast activation in malignancies (‘tumour stroma’) and a series of autoimmune diseases with mesenchymal tissue responses such as immunoglobulin G4 syndrome, systemic sclerosis, myositis, axial spondyloarthritis and RA.1518 These investigations also showed that the fibroblast activation protein inhibitor (FAPI) signal is associated with the accumulation of activated fibroblasts in the affected tissue using histological examination of the affected tissues.15 De Jong suggested that activated fibroblasts play a key role in the development of overt RA.19 A genome-wide transcription profiling and targeted gene expression analyses on synovial biopsies of individuals at risk of RA indicated an increased risk of RA development with increased chemokine (C-X-C motif) ligand 12 (CXCL12) expression and podoplanin staining, supporting the existence of preclinical synovial changes contributed by resident synovial fibroblasts that are associated with RA development. Using FAPI positron emission tomography-CT (PET-CT) to detect activated fibroblasts in vivo, this study sought to investigate whether we can find imaging evidence that synovial fibroblast activation occurs in individuals at risk of developing RA. Furthermore, this study aimed to characterise the association between synovial fibroblast activation and the development of clinical RA as well as clinical and laboratory findings such as joint tenderness and autoantibodies. Methods Patients To explore the transition from preclinical autoimmunity to clinical inflammation, we established the Individuals at Risk for Arthritis Cohort Erlangen (IRACE) cohort in 2011, approved by ethics (333_16 B). This prospective cohort study enrolled individuals positive for ACPAs but with no clinical signs of RA, recruited from the outpatient and inpatient clinics at the University of Erlangen-Nuremberg. All participants provided informed consent. At baseline and follow-up visits, comprehensive data on demographics, body mass index (BMI), age, smoking history, family history, lifestyle and triggers (eg, infections) were collected. For inclusion in our cohort, patients were not allowed to have or have had arthritis and had to be naïve to anti-rheumatic medications. Inclusion in this analysis was restricted to the availability of FAPI-PET, and all cohort participants who were not pregnant or breastfeeding were eligible. Clinical and laboratory assessments IRACE cohort participants are followed up in annual or semi-annual intervals. At each visit, the following clinical assessments are assessed: joint examination for pain and swelling, visual analogue scale (VAS) for patient global and physician global assessment, physical function by health assessment questionnaire (Health Assessment Questionnaire-Disability Index), impact of disease by Rheumatoid Arthritis Impact of Disease and fatigue by Functional Assessment of Chronic Illness Therapy - Fatigue. C-reactive protein (CRP) level and RF were assessed by turbidimetry (Optilite Rheumafaktor Kit and Optilite C-Reactive Protein Kit, The Binding Site Group Ltd, UK). ACPA levels were measured by ELISA (Euroimmun Anti-CCP Elisa, EUROIMMUN Medizinische Labordiagnostika AG, Germany). 68-Gallium-labelled fibroblast activation protein inhibitor-46 positron emission tomography-CT IRACE cohort participants were invited to FAPI PET-CT scans as slots became available. 68Ga-FAPI-46 PET-CT scans were conducted with a Biograph Vision 600 scanner (Siemens Healthineers, Erlangen, Germany). Patients received a median dose of 153 (SD 55) MBq of 68Ga-FAPI-46, with a median body weight of 77 kg. The median time from injection to imaging was 48 (SD 27) min. Scans covered the region from orbitae to the toes, with an additional separate imaging of the hands (3 min per bed position). PET data were corrected for random and scattered coincidences, decay and attenuation. For attenuation correction, a low-dose CT over the same body area was used (100 kV, 40 ref mAs). All corrections and reconstructions were performed using the PET-CT manufacturer’s software. All data were analysed using Syngo.via (Siemens Molecular Imaging, Hoffman Estates, IL, USA). PET, CT and fused images were evaluated by an experienced nuclear medicine physician with >5 years of experience in reading FAPI PET scans. After the FAPI PET-CT, the patients were followed for RA development. RA was defined by the 2010 European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) criteria.20 68-Gallium-labelled fibroblast activation protein inhibitor-46 positron emission tomography-CT A circular volume of interest (VOI) was placed around the joints showing FAPI uptake above the background to determine manually the mean standardised uptake value (SUVmean) and the metabolic active volume (MAV), indicating the mean radioactive concentration in the target as well as the volume of VOI above the background. The SUV is a dimensionless quantity within a given VOI calculated using (i) the local concentration of radioactivity within the VOI (measured in kBq/mL), (ii) the decay-corrected amount of injected radioactivity (measured in kBq) and (iii) the body weight of the patient (measured in grams).21 There are several summary methods of SUV; the most commonly used are SUVmax and SUVmean, where SUVmax provides a measurement of the tracer uptake within the single hottest voxel, and SUVmean provides a measurement of average intensity of all voxels within the VOI.22 The MAV is a three-dimensional measurement of the volume of the VOI in mL, indicating the biological size of the lesion of interest. From these parameters, each lesion’s 68Ga- FAPI-46 uptake was calculated by multiplying the respective SUVmean and MAV, yielding the total lesion FAPI (TL-FAPI). The TL-FAPI therefore indicates a measurement, where both the concentration of the radiopharmaceutical and the size of the lesion of interest are regarded as biological relevant parameters. These analyses were conducted both at the joint (SUVmean, MAV and TL-FAPI) and patient (total SUVmean, total MAV and TL-FAPI) levels by simple summation of the joint-level quantities. In addition, a FAPI joint index was used to compare intra-articular tracer uptake with blood pool and gluteal muscle as background measurements (determined through a circular VOI with a diameter of 1 cm) as described by Luo et al.18 In this index, any joint/enthesial lesion with an uptake higher than or equal to the blood pool but lower than the muscle uptake was given 1 point. Any lesion with an uptake higher than the muscle but lower than or equal to two times the muscle uptake was assigned 2 points. Any lesion with an uptake greater than two times the muscle uptake was assigned 3 points. This provides an overall semi-quantitative assessment of joint FAPI uptake, which also incorporates the background activity as an anchor. Patients were considered positive for FAPI PET-CT if any joint displayed a significant uptake higher than the background, determined through an uptake equal to or higher than the blood pool (PET-CT joint index ≥1). An individual scan site was considered to show any uptake if the SUVmean was >0. Evaluated sites included the metacarpophalangeal/metatarsophalangeal and proximal interphalangeal joints of the hands and feet, shoulder, elbow, wrist, hip, knee and ankle joints, as well as the biceps tendons and the flexor and extensor tendon regions of the hands. Statistical analysis Characteristics of participants were summarised using means, SD, medians and first to third quartile intervals for continuous or count data and as counts and proportions for categorical data. We undertook between-group comparisons to compare participants who did versus did not develop RA and by the presence of FAPI uptake. Wilcoxon’s rank-sum test was used to compare continuous variables, and Fisher’s exact test was used for categorical variables. Several univariable regression analyses were conducted to explore the association of absolute quantities for FAPI uptake (ie, SUV, MAV, TL-FAPI and TJI) with participant characteristics and the risk of RA development. Since four different FAPI quantities were showing a variable range of values, we re-scaled each measurement in regression analyses to have a mean of 0 and an SD of 1. Therefore, effect estimates where FAPI measurements are predictors correspond to a relative risk or difference associated with 1 SD increase of the FAPI value from its mean. Cox regression was used to examine the relative risk of developing RA over time. For this analysis, the cohort follow-up started on the date of FAPI PET and ended on the date of clinical RA development or the last available visit when clinical RA was not observed. The association between FAPI uptake and a painful joint examination at the joint level was evaluated using a mixed-effects generalised linear model with a log-binomial link function. This model calculates the risk ratio (RR) of a joint being painful for each SD increase of the uptake value from its mean while taking into account the within-participant correlation through a random-intercept term. Ordinary least-squares linear regression was used to examine the associations between patient-level FAPI uptake summaries and clinical and laboratory examinations. The R-squared, Akaike information criterion (AIC) and concordance (Harrel’s C) statistics were calculated for each respective model to examine which quantification of FAPI uptake shows the best fit to the observed characteristics/outcomes. Complete cases were included in each analysis without missing data imputation. Two-sided p values <0.05 were considered significant without adjustment for multiplicity. The analyses were conducted using the open-source R software version 4.3.3 (R Foundation for Statistical Computing, Vienna, Austria). Results Baseline characteristics of the participants Among the 130 participants in our RA at-risk cohort, FAPI-PET was available for 18 patients who were positive for ACPAs and clinically suspect arthralgia who had no joint swelling and did not fulfil the ACR/EULAR 2010 Criteria for the Classification of Rheumatoid Arthritis. Participants underwent 68Ga-FAPI-46 PET-CT between July 2019 and November 2024, and the median follow-up after FAPI PET was 42 (IQR 13–211)weeks. The mean age (SD) was 47 (14) years. Eleven participants (61%) were women. Baseline characteristics are summarised in detail on table 1. Notably, CRP levels increased most of the patients at an average value of 10 (14) mg/dL (normal value <5 mg/dL). The mean ACPA value at baseline was 103 IU/mL (143). Eleven participants (61%) were additionally positive for RF. The mean number of tender joints was 1.44 (2.68). Patients reported an average VAS global disease activity of 32/100 mm (30), whereas physicians reported an average VAS of 11 (14). Table 1Baseline characteristicsCharacteristicOverall, N=18*With RA,n=5*Without RA,n=13*P value†Age47 (14)50 (14)46 (14)0.3Sex0.3 Female11 (61%)2 (40%)9 (69%) Male7 (39%)3 (60%)4 (31%)Smoking>0.9 Current smoker6 (33%)2 (40%)4 (31%) Ex-smoker5 (28%)1 (20%)4 (31%) Never smoked7 (39%)2 (40%)5 (38%)Follow-up duration, weeks42 (13, 211)12 (11, 20)97 (40, 211)N/AACPA (anti-CCP2, IU/mL)103 (143)206 (233)63 (69)0.059ACPA (anti-MCV, IU/mL)327 (502)342 (203)321 (586)0.2Rheumatoid factor, IU/mL67 (94)117 (148)48 (62)0.3Rheumatoid factor (binary)>0.9 Negative7 (39%)2 (40%)5 (38%) Positive11 (61%)3 (60%)8 (62%)Erythrocyte sedimentation rate, 1 h17 (12)18 (15)16 (12)0.8C-reactive protein level (mg/L)10 (14)18 (25)7 (4)0.5Tender joint count1.44 (2.68)1.20 (1.64)1.54 (3.04)0.7Physician global assessment (VAS)11 (14)12 (13)11 (14)0.6Patient global assessment (VAS)32 (30)33 (37)32 (28)0.8Total mean standardised uptake value12 (12)17 (18)10 (9)0.5Total metabolic active volume27 (37)62 (58)14 (13)0.14TL-FAPI473 (823)1189 (1,354)197 (227)0.093Total FAPI joint index9 (8)13 (11)8 (7)0.4*Mean (SD); n (%); median (Q1, Q3).†Wilcoxon rank-sum test; Fisher’s exact test; Wilcoxon rank-sum exact test.ACPA, anti-citrullinated protein antibodies; CCP2, cyclic citrullinated peptide generation 2 ; FAPI, fibroblast activation protein inhibitor; MCV, mutated citrullinated vimentin; TL, total lesion uptake; VAS, visual analogue scale. Baseline characteristics Mean (SD); n (%); median (Q1, Q3). Wilcoxon rank-sum test; Fisher’s exact test; Wilcoxon rank-sum exact test. ACPA, anti-citrullinated protein antibodies; CCP2, cyclic citrullinated peptide generation 2 ; FAPI, fibroblast activation protein inhibitor; MCV, mutated citrullinated vimentin; TL, total lesion uptake; VAS, visual analogue scale. Detection of fibroblast activation protein inhibitor-active lesions in rheumatoid arthritis-at-risk participants A total of 1080 sites were evaluated in the 18 participants. FAPI uptake at any value was detected in 64 sites; detected most commonly in the knee joints (15 FAPI-positive joints out of 36 examined joints in total), MTP1 joints (15 FAPI-positive joints out of 36 examined joints in total), followed by the wrists and talocalcaneal (subtalar) joints (6 FAPI-positive joints out of 36 examined joints in total). These joints also showed the highest mean MAV, SUVmean and TL-FAPI values. A detailed description of FAPI uptake by joint site is presented in online supplemental table 1. Only three patients had a total FAPI joint index of 0, which could be described as no meaningful uptake above the background, whereas 15 patients showed uptake in at least one site (figure 1) as defined by a non-zero total joint index. Patient-level FAPI uptake summaries are presented in table 1. Examples for 68Ga-FAPI tracer accumulation. 68Ga-FAPI tracer accumulation is visualised by positron emission tomography (PET)-CT (PET-CT). Tracer accumulation in the lateral part of the right knee (A), left shoulder (B) and right ankle (C). Tracer accumulations are marked by red arrows. Left are PET images, centre are PET-CT fusion images and right are CT images. 8Ga-FAPI, 68gallium-labelled fibroblast activation protein inhibitor. Relationship between fibroblast activation protein inhibitor tracer accumulation and joint tenderness Overall, there were 26 tender joints in eight individuals, most commonly in the tarsal, MCP2, shoulder and ankle regions. FAPI uptake was evaluated in 23 of these joints (three of the tender joints were distal interphalangeal joints that were not evaluated in the scan). Among the 64 sites that showed FAPI uptake, 3 (4.69%) were tender, whereas among the 1016 sites that showed no FAPI uptake, 20 (1.97%) were tender, that is, in crude terms, sites showing any FAPI uptake were twice as likely to be tender than those showing no uptake. Log-binomial regression analyses indicate that the TL-FAPI uptake showed the best fit to the tender joint distribution. One SD increase in the TL-FAPI value above the mean was associated with a 14% increase in the probability of a joint being tender (RR 1.14, 95% CI 1.08 to 1.22, p<0.0001). The mean values for FAPI parameters in tender and non-tender joints are summarised in table 2, and individual distribution of sites with tenderness and uptake is summarised in online supplemental table 2. Since the small number of tender joints among those with uptake could increase the chances of an outlier resulting in an exaggerated effect size, this analysis was also repeated, ignoring the uptake values and by classifying each joint as having any versus no uptake. This analysis shows an increased RR similar to the crude relative risk we mentioned above, but with loss of precision due to dichotomisation (RR 2.02, 95% CI 0.61 to 6.68, p=0.248). Table 2Joint-level FAPI measurements in tender vs non-tender joints and risk ratios of tenderness by 1 SD increase in the respective FAPI measurementNon-tender, mean (SD)Tender, mean (SD)RR (95% CI)P valueAICSUVmean0.16 (0.78)0.50 (1.42)1.19 (0.94 to 1.50)0.146212MAV0.37 (2.61)1.33 (5.67)1.18 (0.97 to 1.44)0.096212TL-FAPI0.88 (5.81)7.33 (32.86)1.14 (1.08 to 1.22)<0.001209AIC, Akaike information criterion; MAV, metabolically active volume; RR, risk risk; SUV mean, mean standardized uptake value; TL, total lesion uptake. Joint-level FAPI measurements in tender vs non-tender joints and risk ratios of tenderness by 1 SD increase in the respective FAPI measurement AIC, Akaike information criterion; MAV, metabolically active volume; RR, risk risk; SUV mean, mean standardized uptake value; TL, total lesion uptake. Relationship between fibroblast activation protein inhibitor tracer accumulation, pain and anti-citrullinated protein antibodies by level FAPI uptake in the joints was summarised at the patient level as total MAV, total SUVmean and TL-FAPI as well as the FAPI joint index. These patient-level FAPI summaries were associated with clinical and laboratory findings (figure 2 and online supplemental table 3). When ranked by the coefficient of determination (r-squared of the linear regression model), the strongest association was found between TL-FAPI and ACPA levels, that is, 1 SD increase in the TL-FAPI was associated with 111 IU/mL higher ACPA levels (95%CI 63 to 159, p<0.001, r-squared=0.6). Still, it is necessary to state that this association was likely driven by a few participants with high ACPA levels, since the log-transformation of ACPA levels did not show a significant association. Correlation of FAPI PET/CT parameters with clinical and demographic data. Scatter plots showing selected correlation analyses between FAPI PET-CT parameters; total SUVmean, total MAV, TL-FAPI, total joint index (TJI) and clinical/demographic variables; ACPA levels, age, VAS patient and VAS physician described in online supplemental table 3. Each point represents an individual subject. Blue lines represent linear regression fits with 95% CIs (shaded areas). Correlation coefficients (R) and p-values are indicated in each plot. ACPA, anti-citrullinated protein antibodies; FAPI, fibroblast activation protein inhibitor; MAV, metabolically active volume; PET-CT, positron emission tomography-computed tomography (PET-CT); SUVmean, standardised uptake mean value; TL-FAPI, total lesion FAPI-uptake. Table 3Patient-level FAPI uptake and the relative risk for RA developmentFAPI parameterHR (95% CI)P valueConcordanceAICTotal SUVmean1.92 (0.75 to 4.92)0.1720.6127.1Total MAV2.20 (1.16 to 4.16)0.0160.7423.8TL-FAPI4.58 (1.38 to 15.22)0.0130.7420.3Total joint index1.80 (0.70 to 4.65)0.2250.6327.4AIC, Akaike information criterion; FAPI, fibroblast activation protein inhibitor; MAV, metabolically active volume; SUV, standardised uptake value; TL, total lesion uptake. Patient-level FAPI uptake and the relative risk for RA development AIC, Akaike information criterion; FAPI, fibroblast activation protein inhibitor; MAV, metabolically active volume; SUV, standardised uptake value; TL, total lesion uptake. Although a higher FAPI uptake of an individual joint was associated with a higher probability of that joint being tender, this association did not translate to a particularly strong association of total tender joint counts and FAPI uptake at the patient level. FAPI uptake at the patient level explained approximately 12% of the variability in the tender joint counts at best (respective r-squared values ranged from 0 to 0.12; online supplemental table 3). FAPI uptake and the risk of developing clinical rheumatoid arthritis During the median follow-up of 42 (IQR 13–211) weeks after the scan, five participants developed RA after a median time of 12 weeks. These participants were older on average and had numerically higher ACPA and RF levels, higher CRP and higher total FAPI uptake at the patient level. None of these differences were statistically significant at an alpha level of 0.05 (table 1). The distribution of FAPI uptake across joints by RA development status is summarised in online supplemental table 4. Survival analyses indicated that the risk of RA development over time increases with higher FAPI uptake, measured as total MAV and TL-FAPI. The HRs also indicated an increased risk of RA development with increasing total SUVmean values and total joint index; however, the concordance statistics and AIC indicate that these did not show as strong an association as total MAV and TL-FAPI. The measurement showing best concordance and AIC statistics was TL-FAPI with an HR of 4.58 (95% CI 1.38 to 15.22, p=0.013), as summarised in table 3. Discussion FAP-targeted tracers have been developed for human use, which allow precise visualisation of the distribution of activated fibroblasts in the body.23 These PET tracers have made it possible to quantify mesenchymal tissue activation in the context of cancer and inflammatory disease.24 25 This cohort study shows, as a proof of concept, that FAPI uptake in the joints of individuals at risk of RA is associated with an increased risk of developing RA. This study shows that fibroblast activation precedes RA onset and increases the risk to develop clinical disease. Our findings also constitute in vivo evidence supporting the findings from de Jong et al that signs of mesenchymal-cell activation reflected by increased CXCL12 and podoplanin expression could be related to early fibroblast-like synoviocyte activation as reflected by increased FAPI uptake.19 Our findings suggest that increased FAPI uptake in the joints of at-risk individuals is associated with higher ACPA titres and, higher patient-reported global disease activity, pointing at an interplay between symptom burden, autoantibodies and fibroblast activation in the joints. ACPAs, which likely develop before pain emerges and FAPI uptake in the joints starts, have been shown to induce pain and fibroblast activation.26 27 Additionally, higher ACPA levels and pain have already been described to influence the risk of RA development.28 Notably, that tender joints are more likely to be FAPI-positive than non-tender joints suggests the existence of a link between fibroblast activation and pain. Recent data have shown that synovial fibroblasts can express genes associated with pain that enhance the growth of pain-sensing neurons.29 The association of FAPI uptake with tenderness provides in vivo support to these in vitro findings. We have observed a correlation between age and FAPI uptake, namely, SUVmean and total joint index. To the best of our knowledge, no study has investigated age-related FAPI uptake in joints or entheses. Imaging and molecular data pointed out that activated fibroblasts also play a role in the process of osteoarthritis, which can be visualised through PET in vivo.30 31 As the prevalence of osteoarthritis increases with age, early degenerative changes in the joint could be an alternative source for FAPI signals and might explain the correlation between FAPI signal and age in our analysis. 68Ga-FAPI PET-CT has been applied in various inflammatory diseases to detect fibroblast activation. Luo et al demonstrated that the FAPI signal in joints of patients with established RA correlated with the FDG signal, suggesting that inflammation and fibroblast activation concur in active RA.18 Notably, discordant findings between FAPI and FDG signals have been described, with FAPI-/FDG+lesions being dominated by a leukocyte-rich infiltrate and FAPI+/FDG- lesions dominated by a fibrotic response,15 clearly showing that the FAPI signal is not an inflammation but a resident tissue response signal. A high FAPI uptake in the joints has also been shown to predict response to disease-modifying antirheumatic drug therapy in patients with RA, 32 supporting the disease relevance of synovial fibroblast activation in RA. Similar findings also pertained to patients with refractory RA, in whom FAPI-tracer accumulation decreased in the affected joints, when successfully treated with T-cell engagers.33 Although all these data refer to established disease, no data on FAPI-tracer uptake in at-risk patients have been reported. The observations reported herein are remarkable, as they show that the activation of synovial fibroblasts is not only observed from the beginning of the disease but also influences the trajectory from transition to the at-risk state to the clinical state of the disease. This concept for RA is paralleled by our findings in patients with psoriasis where increased FAPI uptake predicted development of psoriatic arthritis.34 Notably, since fibroblast activation has been shown to be relevant for triggering bone damage in RA by expressing RANKL (Receptor Activator of NF-kappaB Ligand) and inducing osteoclast formation, these data offer insights to explain the decline of volumetric peripheral bone mineral density and the formation of cortical microlesions (cortical microchannels) in ACPA-positive at-risk participants,35 which are associated with RA development.36 The study has some limitations. The study sample size was limited, which limits the accuracy and precision of the estimates. The sample size essentially depended on the availability of FAPI-PET, given the existing demand for tumour evaluation and assessment of primary fibrotic disorders and inflammatory conditions taking precedence over individuals at risk of RA. Despite this small sample size, we were able to detect an overall predictive potential for FAPI for the risk of RA development. Our type 1 error rate is inflated owing to the lack of adjustment for multiple testing. Therefore, this study should be considered as an exploratory risk factor study, and a confirmatory risk factor study with a larger sample must be conducted to establish whether FAPI is an independent predictor of RA risk after accounting for the established risk factors. We did not use data from further imaging such as MRI or ultrasound to examine in detail the correlates of FAPI uptake in individual joints. Another limitation is the non-uniform association of different FAPI uptake quantities with other clinical and laboratory findings. The association of age and FAPI uptake in this context suggests that osteoarthritis could be an important confounder to be taken into consideration in further studies. However, sample size restrictions did not allow us to attempt such adjustments at this time. As a final extension of the sample size limitation, the joint tenderness association is not strong, and the ACPA association is not robust to challenge. However, given the overall association with the risk of RA development, this respective weakness and fragility could also be a sign that FAPI uptake represents a separate dimension of RA onset risk not well accounted for by these well-known risk factors, which argues for the need to further pursue its predictive potential among other known predictors. In summary, significant FAPI uptake is found in the joints of individuals at risk for RA, and the level of FAPI uptake is associated with the risk of RA development. The relationship between FAPI uptake, joint tenderness and ACPA level also supports the notion of fibroblast activation as an early step in RA development; however, our findings on these associations are fragile and need to be re-examined. With the development of preventive interventions in at-risk patients810 these insights become increasingly important, as they offer the potential to improve risk assessment to identify patients that are likely to develop RA within a short period. Supplementary material