Rheumatology · Medical Literature
Therapeutic approaches for difficult-to-treat rheumatoid arthritis: a systematic literature review of current evidence.
Abstract
This systematic literature review (SLR) aims to update the evidence regarding therapeutic strategies in difficult-to-treat rheumatoid arthritis (D2T RA), building on the previous SLR informing the European Alliance of Associations for Rheumatology (EULAR) points to consider for the management of D2T RA. Three research questions addressed efficacy or safety of treatments in patients with RA with (1) active disease and limited treatment options; (2) active disease with ≥2 prior biologic/targeted synthetic disease-modifying antirheumatic drugs (b/tsDMARDs) and (3) poor health-related quality of life and low objective disease activity (non-pharmacological interventions). MEDLINE, Embase and Cochrane Library were searched until July 2025. Meta-analysis and meta-regression were conducted. We screened 8589 records and included 131 studies. For research question (RQ)1, evidence on DMARD efficacy and safety was synthesised across a wide spectrum of comorbidities including obesity, cardiovascular disease, history of malignancy and respiratory comorbidities. For RQ2, all b/tsDMARDs except otilimab demonstrated better efficacy than placebo (mostly high risk of bias). Meta-regression showed efficacy was maintained for Janus kinase inhibitors (JAKi) despite increasing number of prior bDMARD failures (1 to ≥3). Safety results confirmed the increased occurrence of infection and malignancy with JAKi. For RQ3, limited evidence suggested orthopaedic surgical intervention as a potential non-pharmacological option in patients with D2T RA. This SLR summarised evidence supporting DMARD efficacy/safety across multiple comorbidities. In patients with active RA with ≥2 prior bDMARDs, JAKi offer substantial clinical benefits but require careful risk stratification given cardiovascular and malignancy risks. Furthermore, standardised reporting and dedicated studies on non-pharmacological interventions are urgently needed. CRD42024593584.
Clinical relevance
In patients with difficult-to-treat rheumatoid arthritis (D2T RA), what is the comparative efficacy and safety of pharmacological treatments across different comorbidity contexts, biologic/targeted synthetic DMARD (b/tsDMARD) failure backgrounds, and non-pharmacological interventions for those with poor health-related quality of life despite low objective disease activity? __SNAPSHOT__ {"studyDesign":"Systematic literature review (SLR) with meta-analysis and meta-regression, explicitly intended to update the evidence base informing EULAR points to consider for the management of D2T RA","setting":"Multi-database search (MEDLINE, Embase, Cochrane Library); searches conducted through July 2025; international scope; not restricted to a single institution or country","sampleSize":"131 studies included from 8,589 screened records; individual patient numbers not reported in the available abstract","population":"Adults with rheumatoid arthritis fulfilling criteria for D2T RA, stratified across three research question populations: (RQ1) RA with active disease and limited treatment options due to comorbidities (obesity, cardiovascular disease, history of malignancy, respiratory comorbidities); (RQ2) RA with active disease and failure of ≥2 prior b/tsDMARDs; (RQ3) RA with poor health-related quality of life but low objective disease activity, addressed by non-pharmacological interventions","intervention":"RQ1: DMARDs across a spectrum of comorbid conditions; RQ2: All available b/tsDMARDs (biological and targeted synthetic disease-modifying antirheumatic drugs); RQ3: Non-pharmacological interventions","comparator":"RQ1 and RQ2: Placebo (and active comparators where applicable); RQ3: Not reported in the available abstract","primaryOutcome":"Not explicitly defined as a single primary endpoint; efficacy and safety outcomes were synthesised across research questions (specific outcome measures such as ACR response rates, DAS28 remission, or HRQoL scores are not detailed in the available abstract)","followUp":"Not reported in the available abstract","fundingSource":"Not reported in the available abstract; the prior SLR informing the original EULAR D2T RA points to consider was funded by EULAR — applicability to this update SLR is not confirmed from the abstract"}
Key findings
The SLR screened 8,589 records and included 131 studies, organised around three pre-specified research questions. For RQ1 (active disease with limited treatment options), DMARD efficacy and safety evidence was synthesised across a broad range of clinically relevant comorbidities, including obesity, cardiovascular disease, history of malignancy, and respiratory conditions; specific effect sizes and p-values are not reported in the available abstract. For RQ2 (active disease with ≥2 prior b/tsDMARD failures), meta-analysis demonstrated that all b/tsDMARDs except otilimab showed better efficacy than placebo; the abstract notes that the evidence base was predominantly characterised by high risk of bias. Meta-regression analyses were additionally conducted, though specific meta-regression outputs, coefficients, confidence intervals, or p-values are not reported in the available abstract. For RQ3 (poor HRQoL with low objective disease activity), evidence on non-pharmacological interventions was synthesised; results are not described in the truncated abstract. No NNT or NNH values are calculable from the information available. The finding that otilimab failed to demonstrate superiority over placebo in this heavily pre-treated population is consistent with emerging evidence from the ContRAst programme in broader RA populations, where otilimab underperformed active comparators including tofacitinib and sarilumab, though direct cross-study comparisons are not valid.
Practice implications
This SLR establishes that, among patients with ≥2 prior b/tsDMARD failures — the population most closely approximating true refractory D2T RA — the majority of available b/tsDMARDs retain statistically detectable efficacy over placebo, with the notable exception of otilimab. However, the predominance of high risk-of-bias trial evidence substantially limits confidence in the magnitude and precision of effect estimates. The review does not establish which specific b/tsDMARD is superior in this multiply-refractory population, nor does it resolve the optimal sequencing or mechanism-of-action switching strategy. For RQ1, the synthesis of DMARD safety across specific comorbid conditions (obesity, cardiovascular disease, malignancy history, respiratory disease) is clinically important, as these comorbidities are prevalent in D2T RA and are frequently cited reasons for limited treatment options; however, the abstract does not permit determination of whether any drug was specifically favoured or contraindicated in a given comorbid context. For RQ3, the inclusion of non-pharmacological interventions for patients with poor HRQoL despite low objective inflammation is conceptually important given the established heterogeneity of D2T RA — particularly the distinction between persistent inflammatory and non-inflammatory phenotypes — but findings cannot be evaluated from the abstract alone. This SLR is explicitly positioned as an update to the evidence base for the EULAR D2T RA points to consider; as such, its primary utility is in policy and guideline revision rather than direct practice change at the individual clinician level. Results are most applicable to patients meeting established EULAR criteria for D2T RA. Patients with D2T RA due to non-inflammatory drivers (e.g., central sensitisation, fibromyalgia overlay) may be underrepresented in pharmacological trial populations. IMPORTANT LIMITATION: This summary is based solely on the available abstract; full data tables, forest plots, meta-regression outputs, and subgroup analyses required for complete clinical appraisal are not accessible. __LIMITATIONS__ Only the abstract is available; effect sizes, confidence intervals, p-values, and individual study-level data cannot be independently verified or fully assessed.,The abstract explicitly notes that studies informing RQ2 (≥2 prior b/tsDMARD failures) were predominantly at high risk of bias, which fundamentally limits confidence in pooled efficacy estimates.,Inclusion of 131 heterogeneous studies across three very different research questions raises substantial concerns about clinical and methodological heterogeneity within pooled analyses.,The placebo-controlled trial design dominant in RQ2 may not reflect real-world treatment sequences, switching rationales, or the true degree of treatment refractoriness in clinical D2T RA populations.,Specific comorbidity subgroup analyses for RQ1 are likely limited by small numbers of studies per comorbidity category and potential confounding inherent in observational or post-hoc subgroup data.,Non-pharmacological intervention evidence for RQ3 is not described in the abstract; this domain has historically been supported by low-certainty evidence in D2T RA.,Publication bias is a consistent concern in systematic reviews of b/tsDMARD trials, where industry-sponsored studies with positive results are disproportionately published.,The EULAR D2T RA definition requires failure of ≥2 b/tsDMARDs with different mechanisms of action; included trials may use varying eligibility criteria that do not precisely map to this definition.,Generalisability may be limited given the predominance of academic and trial-enrolled populations, which may not reflect the full comorbidity burden and treatment heterogeneity of routine clinical D2T RA practice.,Funding source of this update SLR is not reported in the abstract, precluding assessment of potential sponsorship bias. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ This updated EULAR-informing systematic review confirms that most b/tsDMARDs retain efficacy over placebo even after failure of ≥2 prior agents, with otilimab being the notable exception — though the evidence base is predominantly high risk of bias and does not resolve optimal agent selection or sequencing in multiply-refractory disease. The synthesis of DMARD safety data across clinically important comorbidities (obesity, cardiovascular disease, malignancy history, respiratory disease) and the inclusion of non-pharmacological interventions for non-inflammatory D2T RA phenotypes represent meaningful additions to the evidence landscape. Clinicians should await the updated EULAR points to consider informed by this SLR before modifying practice, and continue to individualise b/tsDMARD selection based on comorbidity profile, mechanism of action diversity, and shared decision-making.