Early characterization of difficult-to-treat rheumatoid arthritis by suboptimal initial management: a multicentre cohort study

Abstract

Objectives

To demonstrate that unsuccessful treatment optimization in early disease is associated with difficult-to-treat RA (D2T-RA).

Methods

In this retrospective multicentre cohort study conducted from 09/2021–03/2022, we enrolled individuals fulfilling the 2010 ACR/EULAR RA criteria diagnosed 2000–2019. The outcome was D2T-RA by the EULAR definition. We used robust regression to examine the associations with delay, dose, duration of methotrexate and discontinuation of glucocorticoids. We tested through multinomial regression which factors were associated with persistent inflammatory refractory RA (PIRRA) or non-inflammatory refractory RA (NIRRA). Sensitivity analysis included a case-control study matching the year of diagnosis.

Results

We enrolled 48 D2T-RA patients and 145 non-D2T-RA controls. Methotrexate was started within 3 months in 16.7% of D2T-RA vs 33.1% of non-D2T-RA (P = 0.011). Adequate duration of methotrexate was obtained in significantly fewer D2T-RA patients (70.8% vs 85.5%). Glucocorticoids were continued beyond 6 months in a higher proportion of D2T-RA patients (70.8% vs 33.8%, P < 0.001). In multiple regression, treatment delay beyond 3 months (OR 0.3; 95% CI 0.1, 0.9) and non-discontinuation of glucocorticoids after 6 months (OR 4.6; 95% CI 2.2, 9.5) were significantly associated with D2T-RA. Treatment delay was significantly associated with PIRRA only, while non-discontinuation of glucocorticoids was significantly associated with PIRRA and NIRRA. Results were replicated in sensitivity analyses.

Conclusion

Failure to start methotrexate within 3 months and not being off glucocorticoids within 6 months are early predictive features of D2T-RA.

Introduction

Difficult-to-treat RA (D2T-RA) is an emerging concept aiming at characterizing a subset of individuals with RA who have active disease despite using several DMARDs. Clinically, D2T-RA is a relevant issue, as it has been estimated that 6–17% of RA patients will eventually present with it in clinics [1].

Recently, a EULAR task force has endorsed a definition of D2T-RA as resistance to multiple biological or targeted synthetic DMARDs with different mechanisms of action along with the persistence of physical symptoms or high disease activity [2, 3]. While the proposed definition of D2T-RA relies on several failed therapeutic trials, the impact of the initial management of RA on the risk of development of D2T-RA has never been assessed.

We have noted that D2T-RA patients had poorly managed RA during the early stages, including hesitancy to start, optimize or maintain MTX and difficulty discontinuing glucocorticoid (GC) therapy. Herein, we hypothesized that the condition of D2T-RA may be showing already within the first year from diagnosis inherently to initial RA management. In other words, we postulate one could early diagnose D2T-RA by assessing treatment-related factors such as failure to discontinue GC and to implement MTX therapy. To demonstrate so, we exploited the EULAR D2T-RA definition in a dataset of RA patients treated according to EULAR recommendations in two different centres with expertise in treating rheumatic diseases.

Patients and methods

Study population and design

The study population was the inception cohort of RA patients diagnosed and followed up at two outpatient clinics in Northern Italy (Padova and Genova). According to the 2010 ACR/EULAR classification criteria, we approached consecutive patients with an established RA diagnosis during routine visits from 15 September 2021 to 15 March 2022. We excluded patients diagnosed before 2000 (i.e. before bDMARDs were broadly available in Italy) and those with a follow-up on therapy of fewer than 36 months, as we deemed it unlikely that a patient would experience failure of conventional synthetic DMARDs (csDMARDs) and two different biological or targeted synthetic DMARDs before that time. We also excluded those who did not receive MTX as the first csDMARD, as EULAR recommends starting with MTX as the first DMARD; in our practice, starting with leflunomide or sulfasalazine has been seldom performed in the initial management of RA. The study was approved by the institutional review board of the University of Padova and conformed to the ethical guidelines of the Declaration of Helsinki as revised in 2000.

Assessment of study outcome

The outcome of this study was the status of D2T-RA. Four expert rheumatologists (M.L., D.A., K.B., A.C.) assessed this status for each patient at the most recent clinical visit and had to be confirmed for two consecutive visits six months apart. To diagnose D2T-RA, we followed the proposed algorithm based on the EULAR points to consider for the management of D2T RA [2, 4]. According to this algorithm, all three of the following requirements had to be fulfilled for classifying participants as D2T-RA:

• Treatment according to EULAR recommendation and failure of ≥2 b/tsDMARDs (with different mechanisms of action), unless restricted by access to treatment due to socioeconomic factors after failing csDMARD therapy (unless contraindicated). If csDMARD treatment was contraindicated, failure of ≥2 b/tsDMARDs with different mechanisms of action was sufficient.

• Participants must have signs suggestive of active/progressive disease, defined as ≥1 of:

  • at least moderate disease activity (according to validated composite measures including joint counts, for example, DAS28-ESR>3.2 or CDAI>10);

  • signs (including acute phase reactants and imaging) and/or symptoms suggestive of active disease (joint related or other);

  • inability to taper glucocorticoid treatment (below 7.5 mg/day prednisone or equivalent);

  • rapid radiographic progression (with or without signs of active disease): change in van der Heijde-modified Sharp score ≥5 points at 1 year; and

  • well-controlled disease according to above standards, but still having RA symptoms that are causing a reduction in quality of life.

• The management of signs and/or symptoms was perceived as problematic by the rheumatologist and/or the patient.

First, the RA diagnosis had to be confirmed. Then, a physical examination was performed, including an assessment of swollen and tender joint count (SJC/TJC), patient global assessment and evaluator global assessment. Such data allowed us to calculate the DAS in 28-joints using CRP (DAS28-CRP) and ESR (DAS28-ESR) or clinical disease activity index (CDAI). We recorded prior and current medications, including current GC therapy and the current dose of prednisone (in milligrams daily). Altogether, such assessments allowed the determination of the status of D2T-RA according to the recently proposed EULAR definition [4]. If there were uncertainties regarding the attribution of the D2T-RA status, any controversy raised by the referring physician was discussed with the first author (A.G.), and an agreement was reached for each patient. Each physician also attributed more prevalent persistent inflammatory (PIRRA) or non-inflammatory (NIRRA) refractory RA to each D2T-RA patient, according to the viewpoint of Buch et al. [5] Again, in case of doubt, the first author (A.G.) decided whether the patient had PIRRA or NIRRA.

Difficult initial RA management

Treatment-related variables of interest collected were: (i) treatment delay defined as MTX started within 3 months, between 12 and 3 months, or >12 months since diagnosis; (ii) maximum tolerated dose of MTX ≥15 mg weekly; (iii) persistence on MTX >6 months; and (iv) persistence on GCs therapy beyond 6 months. Variables were collected after 12 months from the beginning of MTX therapy. We also recorded whether suboptimal use of MTX was associated with intolerance or contraindications.

Other covariates

According to the literature [3, 6–9], the following non-treatment related variables were more frequently reported in patients with refractory/D2T-RA: female sex, younger age at disease onset, seropositivity for RF or ACPA, obesity (defined as a body mass index ≥30 kg/m²), smoking status, treatment delay, erosions (present vs absent according to X-rays of hands and feet; missing radiographs were imputed as no erosions). Covariates were collected by each referring physician, who went through electronic medical records and paper charts of each participant. Study data were anonymised, collected and managed using REDCap electronic data capture tools hosted at the University of Padova [10].

Sensitivity analyses

To address potential confounding from secular trends in RA management, we performed a matched case-control study of D2T-RA and non-D2T-RA patients having the same year of diagnosis. Furthermore, as D2T-RA encloses both inflammatory and non-inflammatory mechanisms, we tested variables significantly related to PIRRA and NIRRA.

Statistical analysis

We conducted univariate analyses to compare the clinical characteristics of D2T-RA and non-D2T-RA groups. We reported frequency variables (categorical) as absolute numbers and relative percentages, while continuous variables, except where defined otherwise, were reported as mean (s.d.). Where appropriate, the associations of interest between the status of D2T-RA and categorical variables were analysed with the χ² test of independence or Fisher's test. All continuous variables were compared with the Mann–Whitney U test as appropriate according to non-normal distributions and unequal samples in the unmatched cohort. Univariable and multivariable robust regression was performed using D2T-RA as the dependent variable and risk factors after one year of DMARDs therapy (age at RA diagnosis and initial management-related factors) as independent variables. To further characterize the association between initial management-related factors and D2T-RA, we used the case-control matching procedure in SPSS to perform a 1:1 matched case-control study of D2T-RA with non-D2T-RA patients having the same year of diagnosis. Finally, we performed multinomial logistic regression with types of DT2 RA (no D2T-RA vs PIRRA vs NIRRA) as the dependent variable. Patients were asked for relevant missing data, and a complete data analysis was performed; smoking status was missing in over 50% of cases and was not analysed. Statistical significance was considered for a value of P < 0.05. All analyses were conducted in IBM SPSS Statistics version 26 (Armonk, NY, USA) and Stata Statistical Software (Release 16.0, StataCorp LL, College Station, TX, USA). GraphPad Prism version 9.0 for Windows (GraphPad Software, San Diego, California, USA) was used to build figures.

Results

Patients’ disposition

We have illustrated patient disposition in Supplementary Fig. S1, available at Rheumatology online. We evaluated 243 consecutive RA patients during the study period (Padova n = 204, Genoa n = 39). No patient had started with leflunomide or sulfasalazine as initial treatment for RA as for exclusion criteria. After excluding those diagnosed before 2000 (n = 41), those diagnosed after 2019 (n = 7), and those who did not fulfil ACR/EULAR criteria for RA (n = 2), there were 193 patients treated according to the ACR/EULAR recommendations. No participant had restricted treatment access to care due to socioeconomic factors as all had equal rights to access the public health system care providers by free national laws. After applying the EULAR D2T-RA definition, we enrolled 48 participants fulfilling D2T-RA criteria and 145 non-D2T RA patients (Supplementary Fig. S1, available at Rheumatology online; Table 1).

Baseline characteristics of D2T-RA compared with non-D2T-RA patients

Univariate and descriptive statistics of the unmatched and matched cohort are presented in Table 1. There were no significant differences between patients set in Padova and those in Genova; hence, the two groups were analysed as one group of individuals with RA. D2T-RA patients were, on average, four years younger at disease diagnosis than non-D2T, with a significantly higher percentage diagnosed under 60 (91.7 vs 74.5%, P = 0.012). Treatment delay was significantly longer in D2T-RA patients, as the proportion of patients who started MTX within three months was twice as low as non DT2. Furthermore, an adequate duration of MTX therapy was established in a lower percentage of D2T-RA patients than the unmatched non-D2T-RA. Both groups reached adequate dosages (at least 15 mg weekly) in just above 50% of the cases, with no differences between DT2-RA and non-D2T-RA patients. Comorbidities were balanced between DT2-RA and non-D2T-RA patients (Table 1). The case-control study found the same differences, except for age at the disease onset (Table 1).

Fulfilment of the EULAR D2T-RA definition

All participants (D2T-RA and non-D2T-RA patients) were treated according to the EULAR recommendations. There was no participant whose csDMARD treatment was contraindicated, and all failed at least one csDMARD. All DT2-RA patients failed more than two b/tsDMARDs (100%) against only 37.2% of non-DT2-RA. The EULAR definition of DT2-RA was fulfilled mostly because of having a moderate-to-high disease activity (97.9%), presenting signs or symptoms of active disease (50.0%) or being unable to taper GC therapy below 7.5 mg daily (22.9%), and less frequently due to rapid radiographic progression (9.3%) or a relevant reduction in quality of life (7.9%) (Table 1). The management of signs and/or symptoms was perceived as problematic by the rheumatologist and/or the patient in all patients classified as D2T-RA and 9% of non-D2T-RA.

Profiling D2T-RA and non-D2T-RA patients

Driven by selection criteria, disease activity and inflammatory markers were higher in D2T-RA compared with non-D2T-RA patients (Table 1). Likewise, D2T-RA individuals were taking a higher GC dose [5 mg/day (4.4; 8.1) in D2T-RA vs 5 mg/day (2.5; 5.0) in non-D2T-RA; P = 0.033] and received a greater number prior b/tsDMARDs [3 (2; 5) vs 2 (1; 3); P < 0.001]. D2T-RA patients also used more opioids (10.4 vs 2.8%, P = 0.044) and NSAIDs (27 vs 9%, P = 0.001), while they used concomitant csDMARDs less frequently (42.9 vs 59.6%, P = 0.040; MTX 61.9 vs 76.1%, P = 0.185) than non-D2T-RA.

Difficult initial RA management is associated with D2T-RA

Univariable logistic regression evidenced that GCs use, treatment delay and persistence on MTX were significantly associated with D2T-RA (Table 2), while age at onset was not. The multivariable model, including such predictors (Table 2), was statistically significant (P < 0.001), explained 27.4% (Nagelkerke R²) of the variance in D2T-RA and correctly classified 76.7% of cases. We obtained an area under the ROC curve (AUROC) of 0.78 (95% CI 0.71, 0.86), which is an acceptable level of discrimination. In the matched case-control study (n = 96), the associations with D2T-RA were maintained only for treatment delay (OR 4.0, 95% CI 1.4, 10.9; P = 0.007) and GCs use (OR 3.2; 95% CI 1.3, 7.8; P = 0.012). The final model with two variables retained good discriminative capacity (AUROC 0.73; 95% CI 0.63, 0.84).

Association of difficult initial management with PIRRA

We analysed the effects of several factors on PIRRA and NIRRA (compared with non-D2T-RA) through multinomial logistic regression (Fig. 1). In NIRRA patients (n = 17/48), only GC use was significantly associated with the outcome (OR 3.7; 95% CI 1.3, 10.8; P = 0.015). Among PIRRA patients (n = 31/48), we found significant associations for GC use (OR 5.3, 95% CI 2.2, 12.9, P < 0.001), but also treatment delay (MTX <3 months vs >12 months: OR 0.2; 95% CI 0.1, 0.8, P = 0.015; MTX 3–12 months vs >12 months: OR 0.4; 95% CI 0.2, 1.1, P = 0.077).

Discussion

This study was the first application of the entire EULAR D2T-RA definition to a real-life inception cohort of RA patients. Providers must carefully assess the requirements of this definition as DT2-RA does not refer to generic patients refractory to several courses of biologics; thus, it can be complex to evaluate the status of D2T-RA outside a clinical study. In this direction, the original finding of our study was that difficult initial RA management could be predictive of D2T-RA.

The factors defining difficult initial RA management were the inability to discontinue GC therapy and the delay in introducing MTX. The introduction of MTX within three months from RA diagnosis was significantly protective against D2T-RA, though it was pursued in a minority of participants. Indeed, the primary reasons for treatment delay were a generalized approach of physicians with glucocorticoid monotherapy in most patients diagnosed with RA instead of starting with MTX soon after recognizing the disease, patients’ fear of MTX therapy, delays due to screening for infections, and longer than expected follow-up visits due to difficulties to get an appointment in a short time-frame.

Nevertheless, delay to initial treatment was the most important modifiable risk factor of refractory RA [OR (95% CI) 1.63 (1.09, 3.06)], as defined by the failure of >3 DMARD courses [6]. The median delay between the onset of symptoms and the beginning of MTX was longer compared with the reported one of a national multicentre, prospective study of RA patients initiating DMARD therapy after 2015 (>2 years vs 6–7 months). However, we included patients diagnosed before 2015, and our data align with those Bécède and colleagues reported [6]. Moreover, the delay in diagnosing the disease was similar in D2T-RA and non-D2T-RA, and it could not have affected the risk of developing D2T-RA [11]. In our cohort, treatment delay may have been associated with increased disease activity before starting MTX therapy, a known risk factor for decreased likelihood of sustained remission [12]. Overall, our results suggest that treatment delay is an important factor in determining D2T-RA due to inflammatory mechanism; this is also supported by the unilateral association of treatment delay with PIRRA and not NIRRA.

Optimized MTX therapy should follow rules that include titrating to the maximum tolerated dosage, using more tolerated formulations such as the subcutaneous route, and GC bridging for the first weeks due to slow onset of action. Notwithstanding EULAR recommendations for managing early RA, real-world data report low rates of optimised MTX therapy [13]. The usual initial dosage in those RA patients starting MTX is 15 mg weekly in Italy [11]. In our cohort, though, many D2T-RA patients did not maintain dosages higher than 15 mg for long, with the high prevalence (90%) of those reporting intolerance to 20 mg weekly suggesting that this was the main issue limiting MTX use [13]. Indeed, we failed to evidence an association between D2T-RA and maximum MTX tolerated dose, probably because very few patients could maintain treatment with dosages >20 mg weekly. Our data also showed an association between short-lived MTX therapy (6 months) and D2T-RA. Indeed, there was a significant association between no concomitant MTX use to b/tsDMARDs and D2T-RA, suggesting that many patients who initially had suboptimal MTX therapy eventually discontinued MTX in favour of b/tsDMARDs monotherapy [14, 15]. However, retention rates of TNF inhibitors with concomitant MTX may be higher than TNFi monotherapy [16, 17].

Another crucial finding of this work is that the persistence of GC therapy after six months was a hallmark of D2T-RA (OR 2–9). EULAR recommends that glucocorticoid bridging be used for 4–6 weeks in the initial management of RA. Still, the minimum or maximum duration of this bridging is unclear, and often, it is continued beyond six or even 24 months [18]. Continuous GC use has been linked more to non-inflammatory mechanisms (i.e. more associated with NIRRA) than active inflammation [19]. Yet, GC therapy was also often indicative of more active baseline disease and the need for progression to biologics [18], and we noted an association between persistence on GC therapy beyond six months and both PIRRA and NIRRA. In our practice, patients newly diagnosed with RA are seen frequently during the first six months, generally every 1–3 months, as this is a critical time frame for achieving the best clinical outcomes. Knowing whether the patient is still on GC therapy six months after diagnosis may be a warning sign that the disease is becoming difficult to treat. While few patients come off GC therapy by three months, assessing GC discontinuation at 12 months could be too late for intervening on the risk for progression to D2T-RA. Therefore, we chose the glucocorticoid therapy cut-off of 6 months as it was the most reasonable and pragmatic choice to prevent D2T-RA. In our cohort, only about 30% of RA patients could discontinue GC after one year, akin to other real-life settings of DMARDs-experienced patients with RA treated with GCs [20]. However, most RA patients categorized as D2T-RA never discontinued GC thereafter. Permanent discontinuation of GC therapy is uncommon in inflammatory arthritis [18, 21], while the use of extended low-dose GCs in RA is associated with uncertainties regarding their role in harbouring subclinical inflammation and increasing adverse events [22, 23].

Our analysis was detailed in many but not comprehensive of all factors contributing to D2T-RA, a complex condition encompassing groups of individuals variably characterized by non-adherence, dissatisfaction with treatments, multiple pain mechanisms, comorbidities, and true persistent inflammatory activity [24]. For instance, we missed information on smoking and level of education. Higher educational attainment has a protective effect on RA risk that could be independent of smoking and obesity [25]. However, the impact of educational level on the risk of D2T-RA is unknown.

Unlike prior reports [6, 24], we did not find an association of D2T-RA status with female sex, probably because the use of different classes of drugs blunted the sex-specific refractory course to TNF inhibitors. Likewise, we also found no association between obesity and D2T-RA. Obese patients tend to respond less to TNF inhibitors, while the response to non-TNF biologics is usually less affected [26]. Again, all D2T-RA patients used at least one non-TNFi class, thus explaining why obesity was not associated with a refractory course of RA.

Limitations of the study

The main limitation of this study was its retrospective design, prone to selection bias, and missing disease activity scores in the first year of MTX therapy. However, medications were used as a proxy for disease activity. Other limitations include the lack of external validation in a different and prospective RA cohort, underestimation of radiographic progression as a criterion for D2T as radiographs were not performed consistently, and under-representation of non-White groups of patients.

Study interpretation and generalizability

This study aims to reduce treatment delay and obtain early discontinuation of GCs in managing RA patients. Furthermore, as such aspects are already detectable after one year of therapy, this study may inform the criteria for selecting D2T-RA patients for clinical trials.

Conclusions

We have provided evidence there are treatment-related, modifiable factors in the first 6–12 months of disease management that could alter the development of DT2-RA.

Supplementary data

Supplementary data are available at Rheumatology online.

Data availability statement

The data underlying this article will be shared on reasonable request to the corresponding author.

Contribution statement

A.G. provided the conception of the study, literature search, data collection and interpretation, drafting of the article, and revised it critically for important intellectual content; M.S., M.L., K.B., A.C. and D.A. provided data collection; A.G., A.D. and M.Z. provided data search, interpretation of data, drafting the article, revised it critically for important intellectual content; M.L., K.B., A.C. and D.A. performed clinical assessments; all authors revised the article critically for important intellectual content and gave final approval of the version to be submitted.

Funding

No author received any financial support or other benefits from commercial sources for this work. This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors.

Disclosure statement: The authors have declared no conflicts of interest.

Acknowledgements

We wish to thank all patients participating in this study.

This research was done without direct patient involvement. Patients were not invited to comment on the study design and were not consulted to develop patient-relevant outcomes or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy.

The study was approved by Comitato Etico per la Sperimentazione (CESC) della Provincia di Padova (protocol number 45140) and conformed to the ethical guidelines of the Declaration of Helsinki as revised in 2000.

The lead author (the manuscript’s guarantor) affirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

References