What to do when the first TNF inhibitor fails in rheumatoid arthritis: stratified expert recommendations from a scoping review and Delphi consensus

Background: Despite widespread use of tumor necrosis factor inhibitors (TNFi) as first-line therapy in rheumatoid arthritis (RA), up to 40% of patients fail initial treatment. Subsequent therapeutic choices remain poorly structured, with limited evidence-based guidance to inform individualized post-TNFi decision-making. Objective: To develop evidence-informed, profile-based recommendations to guide treatment selection after inadequate response to a first TNFi in RA, combining evidence and expert consensus. Design: Delphi-based consensus study informed by a PRISMA-guided scoping review (ScR) and nominal group methodology. Methods: A PRISMA-guided ScR of biologic and targeted synthetic biologic disease-modifying antirheumatic drug (tsDMARDs) after TNFi failure was conducted. Patient profiles were identified by a steering committee, and draft recommendations were evaluated through an anonymized Delphi process. A profile-based decision tree integrated direct and indirect evidence, with evidence strength graded using the Oxford Centre for Evidence-Based Medicine approach. Results: The ScR included 43 studies, mostly exploratory analyses of randomized trials. Scenarios included age ⩾65 years; failure of ⩾2 TNFi; monotherapy; rheumatoid factor/anti-citrullinated peptide antibody status; prominent systemic inflammation; interstitial lung disease (ILD); rheumatoid vasculitis; high cardiovascular (CV) risk or prior CV event; venous thromboembolism (VTE) risk; obesity; high infection risk; osteoporosis; nociplastic pain, depression and fatigue; prior solid cancer; hematologic cancer/lymphoproliferative disease; non-melanoma skin cancer; and pregnancy.Seventeen recommendations were formulated; 15 achieved consensus. Agreed positions included caution with JAK inhibitors (JAKi) in older patients and in those with CV/VTE risk; preference for IL-6 receptor inhibitors or JAKi for monotherapy or prominent systemic inflammation; in RA-ILD, use a b/tsDMARD with a non-TNFi mechanism; rituximab as first choice in rheumatoid vasculitis; abatacept in infection-prone patients; discouraging JAKi in prior malignancy; and TNFi as acceptable during pregnancy. Two statements did not reach consensus: preferential use of non-TNFi in obesity and heightened caution with tofacitinib in osteoporosis or fracture risk. Conclusion: This Delphi-validated, profile-based framework provides a practical tool to support evidence-informed clinical decision-making.