Article Text
PDF
Abstract
Objectives Secondary prevention of rheumatoid arthritis (RA) is generally considered potentially impactful because the entire RA population is believed to experience a symptomatic ‘pre-RA’ phase. We wondered whether this dogma is correct. Therefore we investigated an inception cohort of patients with newly diagnosed RA and studied among them patients who did and did not present with preceding arthralgia at risk for RA.
Methods Consecutively diagnosed patients with RA between 2012 and 2022 were studied (n=699). These patients had either directly presented with clinically apparent arthritis, or had first presented with clinically suspect arthralgia (CSA). Clinical characteristics at symptom onset and RA diagnosis were compared. Whether certain characteristics frequently occurred together was studied using a K-means algorithm after dimension reduction with partial least squares discriminant analysis. To validate that groups differed in long-term outcomes, sustained disease-modifying anti-rheumatic drug-free remission (SDFR) of the groups was studied during a median follow-up of 5.3 years.
Results Patients with RA who had first presented with CSA were younger, more often had a gradual symptom onset and were more often anti-citrullinated protein antibodies (ACPA)-positive. Studying characteristics at symptom onset and RA diagnosis revealed four patient clusters, of which two clusters included almost all patients with a preceding CSA phase. Patients in these two clusters (55% of RA population) were younger, had a gradual symptom onset, longer symptom duration and were more frequently ACPA-positive. Patients with RA in these clusters achieved SDFR less often (HR 0.51 (95% CI 0.37 to 0.68)) than the patients with RA in the two clusters where preceding CSA was infrequent/absent.
Conclusion These data suggest the notion that the entire RA population has an identifiable symptomatic risk stage should be refuted. This may impact on the scope of preventive interventions targeting the symptomatic risk phase.
- Rheumatoid Arthritis
- Risk Factors
- Classification
Data availability statement
Data are available upon reasonable request. Data are available upon reasonable request. Requests for data (such as de-identified participant data) can be made to the corresponding author following publication, and requests will be considered on an individual basis.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The current understanding of rheumatoid arthritis (RA) development suggests that patients with RA experience a symptomatic at-risk phase of arthralgia before the onset of clinical arthritis.
WHAT THIS STUDY ADDS
These data contradict the notion that all patients with RA have a symptomatic at-risk phase, suggesting this only holds true for a subgroup of patients with RA, that is, younger, has a longer symptom duration and slower disease onset and is more frequently anti-citrullinated protein antibodies positive. These patients also have worse long-term outcomes in comparison to patients in a cluster without a clinically suspect arthralgia phase, achieving less sustained disease-modifying anti-rheumatic drug-free remission (SDFR).
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The implication of our findings is that not all patients with RA might benefit in the future from secondary prevention aimed at patients with a symptomatic at-risk phase, but rather a subgroup of patients with RA. However, this subgroup has the worst outcomes and theoretically could experience the most health benefits from earlier treatment.
Introduction
In the last decades, the focus of research in rheumatoid arthritis (RA) has shifted to the phases of RA development that precede the onset of clinically apparent arthritis. This was fuelled by observations that autoimmune processes can become aberrant long before the clinical diagnosis of RA, leading to the assumption that disease processes are less mature in the risk phases and therefore may be better modifiable compared with the phase when clinically apparent arthritis has developed.1 Indeed, several randomised placebo-controlled clinical trials recently showed data suggestive for disease modification when treatment was initiated in the symptomatic risk stage.2–4 A fixed time period with methotrexate or abatacept resulted in a reduced disease burden or lower incidence of RA development, respectively. These proof-of-concept trials showed that secondary prevention is possible, that is, modification of the disease course for individuals at risk for RA by reducing the burden of disease or lowering the frequency of RA development during 2 years follow-up.
With respect to the promising impact, it is generally assumed that all patients with RA have a detectable symptomatic at-risk phase prior to RA development. However, it is unclear whether this dogma is true. Intriguingly, an evaluation of characteristics at inclusion in arthralgia and early arthritis studies shows that the mean age of patients in arthralgia studies is about 8 years younger than that of patients presenting with early RA (figure 1). This age gap cannot solely be explained by identifying patients earlier in the disease course since the median time of progression from arthralgia to clinically apparent arthritis is less than 1 year.5 Based on this observation, we hypothesised that there are subgroups of patients within the total RA population and that patients with identifiable symptomatic pre-arthritis stages belong to a subgroup with a younger age at onset.
Differences in mean age between arthralgia and early rheumatoid arthritis cohorts/studies. Random effects meta-analysis of single means to calculate an overall mean of arthralgia and early rheumatoid arthritis cohorts, the pooled mean estimates are indicated by the red diamond. Graphic produced using Stata V.14. Studies used: CSA Leiden,17 Seropositive arthralgia cohort, Reade Amsterdam,18 SONAR, Rotterdam,19 CSA Rotterdam,17 Leeds ACPA positive arthralgia cohort,20 APIPPRA,4 ARIAA,3 TREAT EARLIER,2 EAC Leiden,21 Early Arthritis Reade Amsterdam,22 tREACH,23 Yorkshire Early Arthritis Register (YEAR),24 ESPOIR,25 CATCH,26 BARFOT,21 ERAS27 (view online supplemental file 1 for further details).
Supplemental material
Whether or not the overall RA population has an identifiable symptomatic risk stage is critical to the potential impact of studies or efforts aimed at secondary prevention, as positive outcomes found will not translate to patients without an identifiable symptomatic risk stage. Therefore, we aimed to assess if the total RA population has a symptomatic at-risk stage preceding the development of clinically apparent inflammatory arthritis. We are uniquely able to study this because of our setting in which efforts are being made to gain early access for both patients with recent-onset arthralgia and early arthritis. We took advantage of a large inception cohort of patients with RA, and among them evaluated the patients with RA who first presented with clinically apparent arthritis and those who had first presented at the outpatient clinic with clinically suspect arthralgia (CSA). This latter group was, in retrospect, in a symptomatic pre-RA phase at that time. We compared characteristics at symptom-onset and at RA diagnosis and searched for patient clusters. This way, we will draw a tentative conclusion about the likelihood that the entire RA population experiences a recognisable symptomatic at-risk phase before clinically apparent arthritis occurs.
Methods
Setting
This study was conducted at the Leiden University Medical Center (LUMC), Leiden, The Netherlands. Our centre has several ways to enable early access to patients suspected of arthritis. Previous studies have shown that delays are mainly caused at the general practitioner (GP) level; efforts are being made to educate GPs and enable early access for patients whose GP suspects they have clinical arthritis.6 One of these efforts is the Early Arthritis Recognition Clinic (EARC), a clinical arthritis screening clinic organised at the rheumatology outpatient clinic for quick referrals from GPs (a ‘1.5 lines screening clinic’). Since 2010, patients in whom GPs suspected but were unsure of the presence of clinical arthritis are referred to this screening clinic. At the EARC patients filled in a short questionnaire about their joint symptoms, after which they were seen by an experienced rheumatologist (AHM or other senior rheumatologists with expertise in CSA and early RA) who performed a physical joint examination.7 This screening clinic has no waiting list and patients with early arthritis or CSA are evaluated later at the regular outpatient clinic for full work-up by a rheumatologist, whereas patients without these conditions are directly sent back to the GP. Additionally, patients suspected of clinical arthritis who are directly referred to the outpatient clinic (and not via the EARC) are generally seen within 2 weeks. All consecutive patients with newly diagnosed early clinical arthritis (either recognised at the EARC or via direct GP referral) are included in Leiden Early Arthritis Cohort (EAC). Furthermore, patients with clinically suspect arthralgia are included in the observational CSA cohort. Since the LUMC is the only referral centre for rheumatic diseases in a healthcare region of about 500 000 inhabitants, this means that access to rheumatology care is facilitated for all patients suspected of (developing) RA visiting their GP in our region. Hence, the EARC is a method to allow early access and its setting is presented in a flow chart (online supplemental figure S2).
Early Arthritis Clinic
From the total Leiden EAC, we studied all patients with RA consecutively included between January 2012 and May 2022.8 RA was defined as a clinical diagnosis plus fulfilling of the 1987 and/or 2010 RA criteria.9 10 In this cohort, patients are prospectively followed up at 4 months, 12 months and annually thereafter. During visits, physical examinations with joint counts, questionnaires and laboratory measurements were done (serum sample tested for C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), IgG anti-citrullinated protein antibodies (ACPA) ELiA CCP (second generation anti-cyclic citrullinated peptide antibodies assay) Phadia, Nieuwegein and IgM rheumatoid factor (RF; in house ELISA).11 Written informed consent was obtained from all patients; the study was approved by the local institutional review board.
Patient and public involvement statement
Patient partners were involved in the study design of the cohorts.
First presentation with a CSA as pre-RA phase
Part of the patients with RA had presented earlier on to the LUMC with CSA and were included in the CSA cohort, prior to inclusion in the EAC. The diagnosis of CSA is based on the clinical expertise of the rheumatologist. Inclusion criteria for the CSA cohort were: recent-onset (<1 year) arthralgia which was likely to progress to RA based on the clinical expertise of the rheumatologist. By definition, patients with clinical arthritis at physical examination were excluded from taking part, as well as patients with a different more likely cause of their joint symptoms, such as fibromyalgia or osteoarthritis. Since GPs are discouraged from auto-antibody testing prior to referral and patients are included based on the clinical suspicion of RA development of the rheumatologist prior to laboratory testing, inclusion in the CSA is independent of ACPA status. Previous work has demonstrated that the rheumatological expertise accurately distinguishes CSA patients from other patients with musculoskeletal symptoms and that especially the specificity is high (specificity of 93%, OR of 55).12 Fulfilment of the European Alliance of Associations for Rheumatology (EULAR) definition of arthralgia suspicious for progression to RA was not required as this definition was developed in 2016, while the cohort started in 2012.13
Baseline visits in the CSA cohort consisted of physical examination, questionnaires, blood sampling in a similar fashion as in the EAC. Follow-up visits were scheduled at 4, 12 and 24 months.14 Follow-up ended when patients developed clinical arthritis, verified at physical joint examination by rheumatologists, or after completion of the 2-year follow-up period. If patients developed clinical arthritis they were invited to take part in the EAC cohort. Written informed consent was obtained from all patients; the study was approved by the local institutional review board.
The CSA cohort was used to identify patients with a pre-RA stage of arthralgia. To verify this classification, and substantiate whether patients with RA who had presented to the outpatient clinic with joint symptoms before developing clinical arthritis were all in the ‘CSA-yes group’, medical files of all patients with RA were studied. It was evaluated whether patients had presented the year before RA development with CSA-like symptoms while they were not included in the CSA cohort.
Comparisons at RA diagnosis
All analyses comparing patient characteristics were done on patients with a clinical diagnosis of RA who fulfil the 1987 or the 2010 RA criteria. Patients with other inflammatory arthritides such as undifferentiated arthritis, axial spondylarthritis, psoriatic arthritis, osteoarthritis and RS3PE were excluded from our analyses. Patient characteristics were compared between patients with RA who first presented with CSA and those who first presented with clinically apparent arthritis. This concerned clinical characteristics about symptom onset and current symptoms that were collected at the time of RA diagnosis, together with laboratory characteristics such as CRP, ESR, ACPA and IgM RF. Notably, from the patients who presented with CSA, it was certain that they had a symptomatic pre-arthritis stage. However, patients who presented with clinical arthritis either had no symptomatic pre-arthritis phase, or they had such a phase but did not present at the rheumatology clinic at that point in time. In order to tackle this issue, we sought to cluster patients based on their clinical and laboratory characteristics. Hereby we assumed that there are patients without and with a symptomatic pre-RA phase and that the latter group would have shared characteristics at symptom onset or clinical arthritis onset, resulting in patient clusters.
Long-term outcome
To understand whether the differences in diagnosis influenced the long-term outcomes we studied the patients in the identified clusters in relation to achieving sustained disease-modifying anti-rheumatic drug-free remission (SDFR). Follow-up for SDFR differed among patients with RAs, which is inherent to the design of an observational cohort (median follow-up of 5.3 years, IQR 3.4–6.7 years, min 0.03 years-max 9.7 years). SDFR is a proxy to cure in RA; this beneficial outcome is achieved by some but not all patients with RA, it is therefore suitable to differentiate patients. SDFR was defined as absence of clinical synovitis (swollen joints at physical examination) after discontinuation of disease-modifying anti-rheumatic drug (DMARD) use or spontaneous resolution with at least 1 year of follow-up. This means that patients needed to have at least 1 year of remission after DMARD discontinuation in order to achieve this outcome. The median follow-up after achieving DMARD-free remission was 2.9 years (IQR 1.3–4.5 years), this ensured that the DMARD-free remission was truly sustained. If patients had a flare after having achieved DMARD-free remission they were still not included in the SDFR group in our analysis. This sustainability in the definition of SDFR and the long follow-up duration after DMARD-stop allowed for a stringent definition of remission. This is different from, for example, conventional Disease Activity Score (DAS) remission. SDFR is more linked to the individual characteristics of patients rather than the treatment intensity in order to achieve remission. Secondly, a long-term outcome that reflects the subset of patients with a severe disease course was studied, which is, the use of a biological DMARD within the first 5 years of inclusion after RA-diagnosis. All patients with RA were treated in routine care according to the national and international guidelines. Conventional synthetic DMARDs were started, with methotrexate as the first choice. Another conventional synthetic DMARD was started or added if methotrexate failed. Biological DMARDs were considered when patients with RA failed at least two conventional synthetic DMARDs. Moreover, in our centre all indications of biological DMARDs are evaluated in a weekly meeting of rheumatologists before initiation, in order to ascertain careful biological DMARD use. This may have led to biologics being used sparingly, only in the most severely affected patients.
Statistics
First, the patients with RA initially presenting with CSA or with clinical arthritis were compared using χ2, Mann-Whitney U and unpaired t-tests where appropriate using SPSS V.29. Thereafter partial least squares discriminant analysis (PLS-DA) was used in order to explore whether clinical and laboratory characteristics frequently occurred together in patients with CSA as first presenting symptom compared with patients with arthritis as first presenting symptom. PLS-DA is a dimensionality reduction tool that achieves reduction with awareness of class labels; it converts the data to latent factors. The patients were plotted by their latent factor scores. Missing data were rare (3.5% of all values, online supplemental table S3); as PLS-DA analysis uses only cases with complete data, these missing data were imputed using chained equations (m=20 with 100 iterations) from which pooled estimates were calculated.
Then, clusters were identified by a K-means clustering algorithm using the latent factor scores, using R statistical software and the factoextra package. Since K-means clustering is sensitive to outliers, extreme outliers (patients with a Z-score of the latent factor score >5) were not taken into account in the analysis. To determine the optimal number of clusters, a so-called elbow plot was assessed (the curve of the total within-clusters sum of squares computed for different values of k). Lastly, the long-term outcomes of identified clusters were compared by assessing the time to achieving SDFR analysis using Kaplan-Meier estimators and Cox regression using SPSS V.29. We used an alpha level of 0.05 for all statistical tests.
Results
Comparing patients with RA: clinical arthritis versus CSA as first sign of RA
In total 699 patients with RA were consecutively diagnosed, of whom 616 had presented with clinical arthritis to the rheumatologist and 83 had CSA at first presentation. When comparing these two patient groups at the moment of RA diagnosis, patients with CSA as the first presenting symptom had a longer duration of symptoms (232 vs 84 days), more often a gradual onset (84% vs 68%) and lower swollen joint count (4 vs 6) (table 1).
Characteristics at symptom onset and at RA diagnosis, all collected at the time of RA diagnosis, of patients that first presented with CSA or directly with clinically apparent arthritis
Clustering
Two latent factors were extracted in the PLS-DA analysis and patients were plotted for their latent factor scores (figure 2A). Important variables that attributed to the latent factors were age, symptom duration, number of swollen and tender joints, ACPA, RF, DAS-44 and total Health Assessment Questionnaire Disability Index score (online supplemental file 4 and table S4). The two latent factors explained 13.3% of the variance observed in two patient groups. Subsequently, K-means analysis identified clusters of patients based on their latent factor scores; based on the elbow plot it concerned four clusters (online supplemental figure S4). The patient clusters are shown in figure 2B. Comparing figure 2A,B revealed that clusters A and B included patients who initially had presented with CSA. Clusters C and D in contrast, included none or almost none of these patients.
Clustering of patients with RA based on latent factor scores. (A) Clustering of variables in patients with rheumatoid arthritis (RA) based on initial presentation at the outpatient clinic, with either clinical arthritis (blue dot) or clinically suspect arthralgia (CSA) (red dot). Every dot indicates a single individual. These individuals were plotted to their latent factor scores, which represent how strongly each factor is represented in each individual. Dependent variable: Identified CSA phase. Independent variables: age, sex, acute onset, symptom duration, Health Assessment Questionnaire, Visual Analog Scale (VAS): pain, fatigue, general health, localisation of joint pain: symmetry, large/small joints, morning stiffness ≥60 min, smoking, Disease Activity Score, tender joint count-68, swollen joint count-66, C-reactive protein, erythrocyte sedimentation rate, anti-citrullinated protein antibodies status, rheumatoid factor status (B) K-means cluster analysis distinguished four clusters, of which cluster A and B constituted of patients with RA with CSA-like characteristics whereas patients in cluster C and D had characteristics that were quite different from patients with RA with a CSA phase.
The characteristics of the patient in the four clusters are presented in table 2. Cluster A comprised 27% of the RA population; these patients were characterised by a relatively young age at RA diagnosis (mean 49 years), were mostly women (76%), were largely ACPA-positive (80%), had a gradual symptom onset and the longest symptom duration before clinical arthritis development (median 225 days). Of the patients in this cluster, 31% had presented to the outpatient clinic with CSA before developing RA.
Characteristics of clusters detected with K-means
Cluster B comprised 28% of the RA population. Patients in this cluster were slightly older than cluster A, but younger (mean age 55 years) than the patients in clusters C and D. Also these patients were mostly women (80%). About half of the patients were RF or ACPA positive and the number of swollen joints was higher than in cluster A (median 8 vs 3). 9% of the patients in this cluster had presented with CSA before developing RA.
Cluster C comprised 26% of the RA population. These patients were older (mean age 69 years), more often men (62% vs 24% and 20% in clusters A and B, respectively) and more often had an acute/subacute onset of symptoms. The median number of swollen joints was four. The symptom duration at diagnosis was considerably shorter (median 50 days) compared with the previous two clusters. Seven patients had presented with CSA before developing RA; these patients had presented with CSA very shortly before clinical arthritis occurred; the median interval in these patients was 21 days, compared with the median 105 and 55 days for the patients who progressed from CSA to RA in clusters A and B.
Cluster D comprised 19% of the RA population. None of the patients in this cluster had presented with CSA before RA developed. The mean age was 69, similar to the patient in cluster C. However, patients in this cluster typically had a high number of inflamed joints (median 12 swollen joints, 17 tender joints) and were mostly (83%) ACPA-negative.
Verification of the presence of preceding CSA in the medical files revealed that four patients had presented to the outpatient clinic with CSA-like symptoms while they were not included in the CSA cohort. Of these, one patient was assigned to cluster A, two patients to cluster B and, one to cluster C and none to cluster D.
Long-term outcome
Achieving SDFR during a median follow-up of 5.3 years was plotted in for the patients in the four clusters (figure 3). The frequency of SFDR increased in a stepwise manner with the highest rate in cluster D (online supplemental table S5). However, since the curves of clusters A and B, and that of clusters C and D run closely together, they were combined for statistical analyses. Patients in clusters A and B achieved SDFR less frequently than those in clusters C and D (HR 0.51 (95% CI 0.37 to 0.68)). Thus, the clusters with patients without an identifiable symptomatic pre-arthritis stage had a more favourable long-term outcome. Since these patients were also less frequently ACPA-positive, analyses were repeated correcting for ACPA. When corrected for ACPA, SDFR was still achieved less frequently in clusters A and B than in clusters C and D (HR 0.82 (95% CI 0.60 to 1.17)). However, the coefficient of cluster assignment was no longer statistically significant.
Patients in CSA-like clusters develop SDFR less frequently. Above: Kaplan-Meier curves of SDFR development within 5 years follow-up of all clusters identified by K-means analysis. Below: Kaplan-Meier curves comparing patients with CSA-like RA (cluster A and B) with patients with RA with substantially different clinical characteristics (cluster C and D). CSA, clinically suspect arthralgia; RA, rheumatoid arthritis; SDFR, sustained disease-modifying anti-rheumatic drug-free remission; HR, Hazard ratio.
Biological DMARD use within the first 5 years after RA diagnosis was more frequent in clusters A and B in comparison to clusters C and D (A: 19%, B: 20% vs, C: 9%, D: 9%) (online supplemental file 6).
Discussion
Secondary prevention of RA is a timely topic and the dogma is that all patients with RA could theoretically benefit. The underlying assumption is that the entire RA population has a recognisable symptomatic pre-arthritis phase. We questioned this dogma based on the notable and consistent large differences in the age of onset of arthralgia in symptomatic patients in at-risk cohorts and the age of patients with RA in early RA cohorts, pointing out that this assumption might only hold true for a subgroup of patients with RA.
In this study, we showed that patients with a recorded preceding CSA phase have significantly different clinical characteristics at the moment of clinical arthritis detection and differences with respect to symptom onset compared with the general early RA patients. When clustering is performed, two clusters were identified that comprised patients with an identifiable CSA phase, whereas such patients were infrequent or absent in two other clusters. The two clusters of patients with RA with a CSA phase are characterised by being younger, more often women, more often auto-antibody positive and a more gradual onset of disease. Whereas the other two clusters more often have an acute/subacute symptom onset, are older at the time of clinical arthritis occurrence and are more often autoantibody-negative. These findings of patient clusters with different proportions of CSA at first presenting symptom suggest that not the entire RA population has a symptomatic pre-disease phase. This is relevant when considering the possible impact of secondary preventive efforts as potential beneficial effects will not translate to patients without a recognisable symptomatic pre-RA phase, which is up to half of the early RA population studied here.
About one-third of the patients with RA in cluster A had presented with CSA before RA diagnosis. The median interval between the onset of CSA and of clinical arthritis was 15 weeks. The patients with RA in this cluster did not all present with CSA previously, but those without a recorded CSA pre-RA period resembled these patients in terms of distribution of symptoms at symptom onset, number of inflamed joints and frequency of RA-related autoantibodies and increased acute phase reactants. Since these patients also had a long symptom duration before RA diagnosis, it is possible that these patients did have a symptomatic at-risk period but did not yet seek medical attention or were not referred by their GPs during that time.
None of the patients in cluster D had presented with joint symptoms at the outpatient clinic before developing clinical arthritis. This does not rule out that these patients did have a symptomatic at-risk period. However, this apparently was not a clinically recognisable syndrome for rheumatologists or GPs. Moreover, the symptom duration at RA diagnosis was relatively short, which in itself is a hampering factor for diagnosing patients in the arthralgia stage. To some extent, this also applies to patients in cluster C. Seven patients in this cluster did present with CSA before clinical arthritis developed. However, the median interval between CSA and RA was only 3 weeks in these patients. That suggests that patients took advantage of the possibilities of early access and that they could present at the outpatient clinic just before the occurrence of clinical arthritis, but also that the symptomatic pre-RA phase was very short. It could also be discussed that these patients almost had RA at their first presentation but it was unrecognised at the clinical joint examination.
A EULAR study group has identified different phases of RA development, including genetic predisposition, environmental factors, autoantibodies, symptoms and thereafter, the occurrence of clinical arthritis.15 It was already recognised for autoantibodies, smoking and genetic risk that not all patients with RA, by definition, have to pass through all of these phases during disease development. Our results are in line with this notion and for the first time suggest that not all patients may pass through a symptomatic at-risk stage prior to RA development.
It is known that the symptoms in the pre-RA period can be diverse and this could present a limitation in our study.13 CSA is a combination of symptoms and signs that resemble RA, in the absence of clinical arthritis. In our cohort, CSA was diagnosed by expert rheumatologists. The EULAR definition of CSA was not an inclusion criterion as the cohort was designed prior to the publication of the definition. An alternative approach to recognising individuals with symptoms at risk for RA would be to screen for the presence of ACPA-positive musculoskeletal symptoms. In that case, the presence of ACPA increases the risk for RA and symptoms do not need to be specific for CSA/imminent RA. It would be interesting to repeat the analyses as shown here, including this at-risk population. Furthermore, general symptoms like fatigue develop already in the pre-RA phase. It is possible that some of the patients in clusters A and B have had less specific symptoms leading to an under-recognition of CSA by GPs and absence of referral.
A limitation of our study is that inclusion in the CSA cohort was used as a proxy for the presence of a symptomatic pre-RA period. To verify whether there were patients who did present to the outpatient clinic in the pre-RA period but were not included in the CSA-cohort, their electronic health records were checked. Indeed, some patients did present with CSA-like symptoms (and for instance were not willing to participate in the CSA-cohort). However, these cases were rare (four individuals) and would not majorly influence the findings as three out of four patients were assigned to clusters A and B.
Intriguingly, cluster D is quite different from the other clusters, patients are relatively old at diagnosis (median 69 years), they often have a (sub)acute onset of symptoms, present with symmetric polyarthritis, are almost all ACPA negative and, despite the abrupt onset with high inflammation, have good long-term outcomes, as these patients most often achieve SDFR and need biological DMARD less often. Although patients in cluster D were diagnosed and classified as RA, their initial presentation resembles to that of patients with RS3PE; because of the older age, rapid onset and high acute phase reactants, although pitting oedema was missing. Interestingly, SDFR was relatively the highest in cluster D, which shows a great similarity to a cluster of ACPA-negative patients who have previously been described based on a serological risk score by Boeters et al.16 Those patients were older, had a more (sub)acute arthritis presentation, a greater inflammatory burden and were similarly characterised by a relative short-term necessity of DMARD treatment similar to this cluster.
We would like to emphasise it was not our main goal to find clusters within RA. We set out to explore whether the total RA population has a symptomatic pre-arthritis phase, fuelled by the observation that patients in clinically suspect arthralgia cohorts are generally 8 years younger than patients in early RA studies. The main result of this study is that several clusters were found, containing different prevalence of people of an identified pre-arthritis stage. Various clustering techniques exist and it is unknown whether different clustering techniques would have led to different findings. Verification and validation of the identified clusters were not the main goals here and are a topic for future studies. Nevertheless, it is interesting that the identified clusters that did and did not comprise patients with a symptomatic at-risk phase explained the observed age difference in studies of arthralgia and RA in general.
In sum, the current data suggest that not the total RA population passes an identifiable symptomatic at-risk or pre-RA stage. The question is whether this is a disadvantage. Although these patients with RA are less likely to be able to be targeted in secondary prevention efforts, these subsets of RA have a more favourable disease outcome as evidenced by a higher frequency of achieving SDFR. Conversely, based on current data, patients with an identifiable symptomatic pre-RA stage are those with the poorest long-term prognosis and may therefore benefit most from interventions at a pre-arthritis stage.
Data availability statement
Data are available upon reasonable request. Data are available upon reasonable request. Requests for data (such as de-identified participant data) can be made to the corresponding author following publication, and requests will be considered on an individual basis.
Ethics statements
Patient consent for publication
Ethics approval
The research protocol for the Leiden CSA cohort (P11.210) was approved by the medical ethical committee of the Leiden University Medical Center (LUMC). The research protocol for the EAC cohort was approved by the medical ethical committee of the Leiden University Medical Center (LUMC) (B19.008). Participants gave informed consent to participate in the study before taking part.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors AvdH, HWvS, SC and AMPB designed the study. SJHK collected the data. SC analysed the data. AvdH, HWvS and SC interpreted the data and wrote the report. All authors approved the final version of the manuscript and were responsible for the decision to submit the manuscript for publication. SC and AvdH are the guarantors for this manuscript.
Funding This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Starting grant, agreement No 714312), and the Dutch Arthritis Society.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.