Is rheumatoid arthritis always preceded by a symptomatic at-risk phase of arthralgia?

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.

Keywords: Classification; Rheumatoid Arthritis; Risk Factors.

Figure 1. 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, Seropositive arthralgia cohort, Reade Amsterdam, SONAR, Rotterdam, CSA Rotterdam, Leeds ACPA positive arthralgia cohort, APIPPRA, ARIAA, TREAT EARLIER, EAC Leiden, Early Arthritis Reade Amsterdam, tREACH, Yorkshire Early Arthritis Register (YEAR), ESPOIR, CATCH, BARFOT, ERAS (view online supplemental file 1 for further details).

Figure 2. 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.

Figure 3. 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.