Abstract
Objective In 2015, the European Alliance of Associations for Rheumatology (EULAR) published recommendations for the use of imaging for the diagnosis and management of spondyloarthritis (SpA) in clinical practice. These recommendations included the use of ultrasound (US) in patients with psoriatic arthritis (PsA), but the management was not clearly distinguished from that of SpA. We aimed to systematically review the literature on the role of US for the management of PsA, and to propose pragmatic algorithms for its use in clinical practice.
Methods A group of 10 rheumatologists, experienced in imaging and musculoskeletal US, met with the objectives of formulating key questions for a systematic literature review (SLR), appraising the available evidence, and then proposing algorithms on the application of US in suspected or established PsA, based on both the literature and experts’ opinions following a Delphi process.
Results The SLR included 120 articles, most of which focused on the diagnostic process. The elevated number of articles retrieved suggests the interest of rheumatologists in using US in the management of PsA. After a consensual discussion on literature data and expert opinion, the following 3 algorithms were developed to be used in practical situations: suspicion of PsA, management of PsA with good clinical response, and management of PsA with insufficient clinical response.
Conclusion The SLR showed interest by rheumatologists in using US to objectively evaluate PsA for diagnosis and management. We propose 3 practical algorithms to guide its use in the clinical management of patients, from diagnosis to the assessment of treatment response. Further studies are needed to define remission and to assess the ability of US to predict disease severity.
Psoriatic arthritis (PsA) is a chronic inflammatory disease characterized by both articular and musculoskeletal (MSK) features. It is associated with a significant disease burden and functional impairment when diagnosis and treatment are delayed.1-3 As no single diagnostic test exists, diagnosis is based on a combination of clinical, laboratory, and radiological variables, as encompassed by the Classification for Psoriatic Arthritis criteria (CASPAR)4; these include the presence or past history of synovitis, tenosynovitis, enthesitis, bursitis, and psoriasis (PsO). The management of patients with PsA has become more effective in recent years due to the advent of new effective drugs, the publication of recommendations on PsA treatment,2 and the development of clinical criteria to assess disease activity as well as the definition of remission status (eg, minimal disease activity). Moreover, it has been shown that even in PsA, the management of patients based on a treat-to-target approach improves disease control and the likelihood of achieving clinical remission and reducing structural progression and functional impairment.5
In recent years, rheumatologists have increasingly used ultrasound (US) imaging to evaluate the peripheral involvement of patients with PsA6-8 for both diagnostic and management purposes, although there remains no standardized strategy to do this.9 US is seen as a complementary tool to conventional clinical assessment that can be applied contemporaneously in the clinical setting. Once regarded as a highly operator-dependant tool, standardization of the acquisition and reading of images has significantly improved its reliability over the last 10 years. In 2015, the European Alliance of Associations for Rheumatology (EULAR) published the first recommendations for the use of imaging for the diagnosis and management of spondyloarthritis (SpA) in clinical practice, proposing the use of US for the detection of peripheral lesions and for their monitoring.10 These recommendations also included patients with PsA, but the management was not clearly distinguished from patients with SpA, and the value of these recommendations in clinical practice still needs to be assessed.
We report a systematic review of the literature aimed to analyze the role of US for the management of patients with PsA in terms of diagnosis, treatment follow-up, and prognosis, and to propose pragmatic algorithms that include information about lesions and articular sites to be assessed in clinical practice.
METHODS
A group of 10 rheumatologists, experienced in imaging and US in rheumatology and representing 7 European countries (Denmark, France, Germany, Hungary, Italy, Spain, and the United Kingdom), met twice in January 2018 and January 2019 using a 2-round Delphi process with the objectives of (1) formulating the key clinical questions for developing the literature search, (2) identifying and critically appraising the available literature, and (3) proposing pragmatic algorithms on the use of US in suspected or established PsA based on both published evidence and expert opinion. The experts were divided into 3 subgroups, each of them focused on a different research area: (1) diagnosis, (2) assessment of treatment response, and (3) assessment of disease state.
Research questions and literature review. Five research questions were agreed upon: (1) the added value of US for diagnosing peripheral lesions in PsA, (2) the added value of US for permitting an early diagnosis or a differential diagnosis of PsA, (3) the added value in monitoring treated patients, (4) the contribution of US findings in defining remission, and (5) the ability of US to predict the severity of disease including structural progression and flares. For each question, it was proposed to additionally look at the anatomical structures examined (ie, joints, tendons, bursa, peritendon, and enthesis), including their location and number.
A systematic literature review (SLR) was performed following a standardized protocol according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.11,12 Eligibility criteria are detailed in Supplementary Material S1 (available from the authors upon request). The search, performed in December 2018, used Medical Subject Headings (MeSH) terms already validated in a previous SLR on rheumatoid arthritis (RA)13 by an expert librarian and was adapted to PsA (see Supplementary Material S1, available from the authors upon request). The following electronic databases were searched: Cochrane library (2018), MEDLINE (from 1973 to December 2018), and Embase (from 1973 to December 2018). The year 1973 corresponded to the first use of US in rheumatology. The search was limited by language (articles in English), and state of the review was not registered in the PROSPERO database. Due to the coronavirus disease 2019 (COVID-19) pandemic, the analysis was delayed and the SLR was actualized according to the same criteria from January 2019 to December 2021.
Dedicated software was used to manage the records retrieved (Version 6.0.20; Zotero). Two participants (HG and MADA) applied screening criteria to review all identified citations by title and abstract. Subsequently, the selected articles were reviewed in detail by HG and randomly by MADA, and those that did not fulfill inclusion criteria were excluded from the SLR. Finally, a hand search was completed by reviewing the references of the included studies; if some publications were missed, they were also examined and included.
Data collection process and quality assessment. The information from the selected manuscripts were extracted using standard forms, developed and previously used for this purpose.14 We did not apply published quality checklists because of the lack of adapted scores; we attributed a binary score (0 as not satisfactory or unclear, and 1 as clear/satisfactory to some defined variables) to the examined papers and the sum of the scores was used to assess the quality of the published studies. The questions included in this quality index are detailed in the Supplementary Material S1 (available from the authors upon request).
Statistical methods. Descriptive statistics were used to report data. Frequencies and percentages were used for categorical variables. The statistical tests of each study were reported during the data collection process.
Ethics approval and patient consent were not sought for the present study since they were not applicable.
Development of the algorithms. The development of the algorithms was primarily based on the literature evidence of each research area (ie, diagnosis, assessment of treatment response, and assessment of disease state), and then circulated for agreement. However, if this evidence was insufficient, the algorithm was mainly based on expert opinion and required a 100% agreement between experts in the Delphi process consensus. In case of lack of evidence, a research agenda was drafted to propose these questions as priority unmet needs.
RESULTS
Overall, the search yielded 14,770 titles, out of which 124 duplicates were excluded; thus, 14,646 titles and abstracts were screened and 370 full texts were assessed, leading to the final inclusion of 120 manuscripts in the SLR. Additional details are presented in the flow chart (Supplementary Material S2, available from the authors upon request). Of these 120 articles, 62 concerned the diagnostic process (including the description of the so-called elementary lesions), 39 concerned the therapeutic monitoring of PsA (activity assessment and treatment response), 7 concerned the remission state, and 12 the severity of progression (9 focused on structural damage and 3 on the prognosis). For reasons of clarity, we reported only the studies with the highest methodology criteria in the tables (Tables 1-3; Supplementary Material S2). Three algorithms were finally obtained (Figures 1-3). The proposed algorithms and the supporting evidence are presented in the following paragraphs.
Definition of US lesions: main studies comparing PsA, healthy subjects, and patients with PsO.
Definition of US lesions: main studies comparing PsA to differential diagnoses.
Treatment responsiveness: main studies.
Algorithm for suspecting PsA. * Target population: (1) patients with PsO with MSK peripheral symptoms not fulfilling CASPAR, (2) seronegative early IA, (3) patients with inflammatory arthralgia without PsO. ** To be defined. CASPAR: Classification for Psoriatic Arthritis criteria; IA: inflammatory arthritis; MSK: musculoskeletal; PsA: psoriatic arthritis; PsO: psoriasis; US: ultrasound.
Algorithm in PsA after poor clinical treatment response. ** To be defined. cs/bDMARD: conventional synthetic/biologic disease-modifying antirheumatic drug; MSK: musculoskeletal; NSAID: nonsteroidal antiinflammatory drug; PsA: psoriatic arthritis; US: ultrasound.
Algorithm in PsA after good clinical treatment response/remission. ** To be decided. cs/bDMARD: conventional synthetic/biologic disease-modifying antirheumatic drug; MSK: musculoskeletal; NSAID: nonsteroidal antiinflammatory drug; PsA: psoriatic arthritis; US: ultrasound.
1. US use for PsA diagnosis.
• Diagnostic algorithm. Most of the retrieved studies focused on the capability of US to detect specific US lesions (ie, target lesions) in PsA; only a few directly assessed the role of US in establishing the diagnosis of PsA (see “PsA diagnosis” section below). Studies describing the target lesions focused on joints, entheses, tendons, but also on other soft tissues (peritendon, bursae) and nails and skin. Table 1 shows the most frequent US target lesions evaluated in diagnostic studies (additional studies in Supplementary Material S2, available from the authors upon request).
Several studies noted the higher sensitivity of US, as compared to clinical examination, for the detection of joint involvement in PsA, especially synovitis, thereby contributing to the establishment of an earlier diagnosis of PsA.15-19 With the earlier detection of lesions, the time between the onset of symptoms and formal diagnosis of PsA can be shortened, resulting in reduced structural damage and improved patient-related outcomes.1,2,15 There was also evidence that US is a sensitive tool for the detection of subclinical changes in patients considered at-risk because of MSK complaints and/or with personal or familial history of PsO.20-25
Although the absence of relevant MSK US inflammatory lesions necessitates questioning the initial suspicion of PsA, the presence of relevant US inflammatory lesions—which includes the above-mentioned synovitis, enthesitis, bursitis, and dactylitis—also requires additional specific features to support diagnosis of PsA. Figure 1 illustrates this proposal. Relevant inflammation was defined as the presence of an unequivocal inflammatory lesion according to the judgment of the examiner. Enthesitis and dactylitis are part of the specific features to support PsA diagnosis and therefore the US findings of these were studied with particular attention. Other characteristic US patterns, such as other extrasynovial or nail lesions, were suggested in the literature but remained insufficiently validated and needed more evidence.26-29 US synovitis alone was considered not specific enough to confirm a diagnosis of PsA based solely on US findings, since the ability of US to distinguish PsA from other chronic arthritides is not yet validated.28,30,31 In this situation, we encourage the rheumatologist to investigate further etiologies before establishing a final diagnosis.
• Patients with PsO. A larger number of US studies compared patients with PsA to those with PsO, and to healthy controls.18,20,21,23-25,32-40 Almost all studies reported higher prevalence of synovitis, enthesitis, and entheseal thickening in patients with PsA compared to those with PsO and healthy subjects.18,20,21,24,35,38,39 One study compared the prevalence of US entheseal abnormalities in PsO with MSK complaints to the combination of US and clinical findings, and found a prevalence of 45% for US only and 36% with the combination of both.25 Most of studies included only asymptomatic patients with PsO and reported prevalence rates of enthesopathy (ie, enthesis abnormality) ranging from 18% to 33%, with higher enthesopathy scores (Madrid Sonographic Enthesis Index [MASEI], Glasgow Ultrasound Enthesitis Scoring System [GUESS]) than healthy subjects.22,23,32-34,37,41 The interpretation of these results is made difficult by the different definitions of enthesopathy and enthesitis among studies, and because most studies are cross-sectional and a few are cohort studies.
Regarding nail involvement, a few studies explored the value of US, and all reported a good sensitivity and specificity for detecting nail lesions compared to clinical (ie, visual) examination as a reference. The detection of nail plate thickening and irregularities had a sensitivity of 75% and a specificity of 94%, and detection of nail bed thickening had a sensitivity of 95% and a specificity of 97%.21 Nail lesions were more frequent in patients with PsA/PsO than in healthy controls,26,42-45 but the ability of US to discriminate patients with PsA from patients with PsO led to inconsistent results across studies.26,42-45 As data were inconclusive on this aspect, we did not include these features in the proposed algorithm.
• Differential diagnoses. Most of the studies compared patients with a suspicion of PsA to healthy controls, patients with PsO, or patients with other inflammatory arthritides, especially patients with RA (Table 2).
Compared to RA, patients with early PsA had a higher prevalence of extrasynovial lesions, including the involvement of extensor tendons of hands, soft tissue edema, and pulley lesions.30,31,46 In one study, the following 3 PsA-associated features were proposed to distinguish PsA from RA in patients with early arthritis: (1) peri-extensor tendon inflammation at the metacarpophalangeal joint level (present in 54% of patients with PsA vs 3% with RA), (2) enthesitis of the central slip of the extensor finger tendon (25% in early PsA vs 0% in early RA), and (3) soft tissue edema around the flexor tendon of the finger (42% vs 3%, respectively).30 In this study, the detection of these US lesions combined with skin lesions (by dermoscopy) raised the specificity of early PsA diagnosis (vs early RA) to 91%. Comparative studies between established PsA and established RA showed no consensus on specific sites able to discriminate PsA from RA, but all publications pointed out more extrasynovial abnormalities encountered in PsA, in particular entheseal lesions.47-50
One study, focused on Achilles tendon enthesis, compared patients with PsA to those with RA and those with osteoarthritis (OA). Patients with PsA and those with RA presented more inflammatory abnormalities (6-37% in PsA vs 2-26% in RA vs 0-15% in OA), whereas patients with PsA and those with OA presented more frequently enthesophytes (47-57% vs 34% in RA), and patients with OA had a lower prevalence of erosions as compared to those with PsA and those with RA (1-12% in PsA/RA vs 0% in OA).48 Finally, some studies compared patients with PsA and patients with fibromyalgia51-54 and suggested that the most discriminant features were a higher number of inflammatory enthesitis lesions (power Doppler [PD] positive enthesitis; 70% in PsA vs 23% in fibromyalgia) and erosions (20% vs 0%, respectively, P = 0.001).51 The most discriminant sites were the Achilles, quadriceps, and lateral epicondyle entheses.51,52 Therefore, we included the presence of enthesitis and other extrasynovial features in the diagnostic algorithm, as US abnormalities are strongly indicative of PsA diagnosis; although, as mentioned in the algorithm, the characteristic pattern including location needs to be confirmed more precisely by additional studies.
• PsA diagnosis. The majority of studies focused on the US detection of PsA elementary lesions, and not on their diagnostic value (eg, added value for the final diagnosis of PsA). Only 7/62 of the selected diagnostic articles evaluated diagnostic impact, but they often used classification criteria, and not diagnosis in clinical practice (Supplementary Material S2, available from the authors upon request).
• Research agenda. The research agenda included the following: (1) to define a threshold for relevant inflammation, based on the number and type of lesions; and (2) to provide a precise definition of the characteristic pattern(s) of PsA, including locations.
2. US use for PsA treatment response.
• Treatment response algorithms. In the management of patients with PsA, US may add value by detecting the persistence of residual inflammation independent of clinical examination.55,56 The treat-to-target approach, which has been shown to reduce structural damage in patients with PsA, requires tight monitoring, with frequent assessment of disease activity. US may potentially be used to assess residual inflammation, hence defining good (or poor) clinical response.
To address this research question, we distinguished 2 clinical situations: (1) poor clinical response and (2) good clinical response, which are illustrated in Figure 2 and Figure 3, respectively. If clinical response is not satisfactory, the confirmation of US inflammation should lead to an escalation of the ongoing treatment, according to current national or international guidelines.57 However, if no US relevant inflammation is observed, additional causes of nonresponse should be considered, or the rheumatologist may carry out further investigation, such as other imaging and/or laboratory testing.58,59 On the other hand, if clinical response is satisfactory, US examination should be performed to confirm the absence of relevant inflammation. Concomitant good clinical and US responses may eventually allow tapering of treatment.60 In case of persistent US inflammation, when a discrepancy between clinical and US findings is observed, the ongoing treatment should at least be maintained. Several studies addressing this question have shown discrepancies between clinical and US evaluation of disease state, and if that is the case, treatment tapering based on clinical evaluation alone may be followed by a rapid flare.61 The decision of treatment increase may be carefully discussed, since the management of a patient in case of US activity alone is not defined. If necessary, treatment increase or change of treatment would depend on national guidelines and consider the type and location of inflammatory features, as proposed in the algorithm. US may be helpful to define the domains involved, in addition to clinical and biological features, and thus be helpful to define treatment strategy according to current recommendations.57
• Activity assessment. US research on the management of PsA has focused mainly on activity assessment, including the development of composite activity scores. Only a few studies evaluated the role of US in treatment responsiveness (Table 3).
Studies comparing US to clinical examination almost always found that US was more sensitive than clinical assessment for detecting synovitis, enthesitis, or other inflammatory lesions.17,62-66 These results were variable across studies and depended on the sites studied. Three studies focused on early PsA (disease duration < 1-2 years), and highlighted the high prevalence of subclinical synovitis/enthesitis in patients with early PsA, ranging from 25% to 76% of synovitis and 95% of entheseal abnormalities, including the presence of PD signal in 5% to 14%.67-69
Comparison between US and clinical, biological variables found weak-to-moderate correlations (coefficients ranged from 0.16 to 0.51; Supplementary Material S2, available from the authors upon request).70-75 Compared to magnetic resonance imaging (MRI), US had a good sensitivity for detecting inflammatory changes.60,76,77 However, its sensitivity and specificity seemed to be variable depending on the sites and lesions studied, with sensitivity varying from 40% to 100% and specificity from 84% to 99%.76-78
• Treatment responsiveness. There were fewer studies about treatment responsiveness (Table 3), but a recent study found that treatment response according to US was significantly higher in the treatment group compared to placebo, with an adjusted mean change in Global Outcome Measures in Rheumatology (OMERACT)-EULAR Synovitis Score (GLOESS) of −9 (standard error [SE] 0.9) in the secukinumab group vs −6 (SE 0.9) in the placebo group (1-sided P = 0.004). In addition, this study showed that changes in US appeared very early after treatment initiation, even within the first week of treatment.79 All studies reported concomitant significant improvement of US and clinical findings in patients with PsA treated by disease-modifying antirheumatic drugs (DMARDs).55,60,79-86
• US scoring. Published data suggest that GLOESS,87 a composite (B mode and Doppler mode) severity score developed in patients with RA, should also be used in PsA because of the high responsiveness for synovial changes with treatment.79 This scoring system evaluates the severity of synovitis at both joint level and patient level by using the sum of the highest synovitis score of each of the joints involved. Regarding the evaluation of enthesitis, several scores were proposed and the group recommended using the recently developed OMERACT enthesitis score.88 At patient level, the situation is different, since several scoring systems are proposed in the literature and some of them include both entheses and joints in the same scoring system (Supplementary Material S2, available from the authors upon request). Nevertheless, Tom et al described elementary lesions of enthesitis and relevant sites of entheses to be scanned in patients with PsA, which may be useful to create a new scoring system.89 One study was based on a web-ranking exercise assessing the potential sites considered as useful by 20 experienced rheumatologists; this could also be a starting point for a global US score to assess disease activity in patients with PsA.90
• Good clinical response/remission. One of the main challenges in PsA is the definition of good response and of remission using clinical scores. Indeed, there was a lack of standardized definitions of US remission. The 2 main definitions of US remission used either PD score = 0 or PD score ≤ 1 (Supplementary Material S2, available from the authors upon request).60,71-73,91 According to these US thresholds, the percentage of patients in remission compared to clinical remission varied from 17% to 69%.60,71-73,91 Based on the lack of published data, the evaluation of remission by US was included within the algorithm of good clinical response, since we did not find sufficient evidence to create a separate US management algorithm. The latest EULAR recommendations for the therapeutic management of PsA (2019 update) included a new recommendation about patients with PsA in sustained remission, suggesting cautious tapering of DMARDs.2 The added value of US to confirm clinical sustained remission could be of interest, but needs to be determined.10
• Research agenda. The research agenda included the following: (1) to determine the threshold of relevant US inflammation and the threshold of US remission using different reference standards of remission (longitudinal construct validity); (2) to determine the appropriate timepoint for US examination of patients on treatment; and (3) to establish a global US score to assess disease activity.
3. Assessment of disease state.
In addition to diagnosis and treatment response, there are other issues to address, in particular, how to evaluate the disease state, including the assessment of PsA severity, the prediction of flares during follow-up, and the structural progression. We did not include these aspects in the proposed algorithms since the published evidence was scarce on these last points.
Few studies evaluated the predictive value of US for structural damage (Supplementary Material S2, available from the authors upon request), and we did not find any data comparing US findings with quality of life for evaluating the severity of PsA. Concerning the comparison with radiographic progression, 1 study found a moderate correlation between US inflammatory scores and radiograph changes.92
There were only 2 longitudinal studies evaluating structural progression over time and the ability of US to predict the occurrence of progression15,93; other older studies were cross-sectional (Supplementary Material S2, available from the authors upon request). The longitudinal study of El Miedany et al included early PsA and found that both US inflammation at baseline and persistent synovitis at 6 months could predict higher structural damage at 12 months; US grayscale (GS) score ≥ 2 and PD score ≥ 2 at baseline were associated with progression of joint damage, where odds ratio (OR) was 2.6 (95% CI 1.3-2.9) for GS and 2.7 (95% CI 1.1-2.8) for PD.15 Persistence of synovitis/enthesitis was also associated with structural progression. Cross-sectional studies found an association between US structural changes and both MRI and radiograph features, both for peripheral and axial damage.76-78,94,95 An increase of 10 points in MASEI score was associated with higher peripheral joint radiograph damage (OR 1.42, 95% CI 1.15-1.72).94
These studies suggest that US severity assessment in patients with new and established PsA is necessary and may help determine the disease prognosis. However, the sites to examine by US for the evaluation of structural changes (eg, joints, enthesis, and number of sites) still need to be defined.
DISCUSSION
The findings of this SLR support the use of US in the management of patients with PsA. Based on the available data, we were able to propose 3 practical algorithms to guide its use for the clinical management of patients with PsA, mainly for helping the diagnostic evaluation and for assessing treatment response. We identified different situations to propose pragmatic clinical guidance with regard to the suspicion of PsA, management of a patient with PsA with good clinical response, and management in case of insufficient clinical response. The findings highlighted that some specific US lesions should be examined to confirm the diagnosis of PsA, because they can help to distinguish this disease from other rheumatic diseases. However, how many sites should be examined and if all these specific lesions should be sequentially or simultaneously evaluated is still unclear, as well as if we can consider some sites as specific to target. The proposed algorithms are feasible but need to be evaluated by further studies to validate their feasibility and to assess their effect in clinical settings. An additional algorithm specifically for remission could be of interest, but evidence on this aspect was insufficient at this time to propose standardized guidance.
Thresholds for defining good or poor clinical response also need to be further defined, in addition to the frequency of US examination during follow-up. Further studies are also needed to define US ability to predict disease severity, whether for structural damage or for risk of flares. New evidence on these issues would help to answer remaining questions raised in this SLR.
ACKNOWLEDGMENT
We would like to thank Prof. Lene Terslev for her great help in this work.
Footnotes
The project was financially supported by the University of Leeds, which received an unrestricted grant from AbbVie.
MB declares speaker fees from UCB, Amgen, Lilly, and Galapagos; support for travel from UCB; and participation on data safety monitoring board of advisory board with Mylan and UCB. MADA declares speaker fees from AbbVie, Amgen, BMS, Galapagos, Janssen, Eli Lilly, Novartis, and Pfizer; consultancy for AbbVie, Janssen, and Novartis; and grants from Pfizer and Amgen. PE is on the speakers’ bureau for BMS, Boehringer Ingelheim, Galapagos, Gilead, Eli Lilly, and Novartis; received consultant fees from BMS, Boehringer Ingelheim, Galapagos, Gilead, Janssen, Eli Lilly, Novartis, Pfizer, and Samsung; received support for travel from Eli Lilly and Boehringer Ingelheim; received grants from Lilly, Novartis, and Samsung; and declares participation on data safety monitoring board of advisory board with AstraZeneca. HG declares a grant from AbbVie. AI declares honoraria, advisory boards, speakers’ bureau, educational grants, and research support from AbbVie, Alfasigma, BMS, Celgene, Celltrion, Eli Lilly, Galapagos, Gilead, Janssen, MSD, Novartis, Pfizer, Sanofi Genzyme, and SOBI. MØ declares research grants from AbbVie, BMS, Merck, Celgene, and Novartis, and speaker and/or consultancy fees from AbbVie, BMS, Boehringer Ingelheim, Celgene, Eli Lilly, Hospira, Janssen, Merck, Novartis, Novo, Orion, Pfizer, Regeneron, Roche, Sandoz, Sanofi, and UCB. The remaining authors declare no conflicts of interest relevant to this article.
P. Emery and M.A. D’Agostino contributed equally as senior authors.
- Accepted for publication September 1, 2023.
- Copyright © 2024 by the Journal of Rheumatology
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.
- 34.↵
- 35.↵
- 36.
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.
- 44.
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.
- 50.↵
- 51.↵
- 52.↵
- 53.
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.
- 64.
- 65.
- 66.↵
- 67.↵
- 68.
- 69.↵
- 70.↵
- 71.↵
- 72.
- 73.↵
- 74.
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.
- 81.
- 82.
- 83.
- 84.
- 85.
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.
- 97.
- 98.
