To identify clinical features that predict the risk of meeting difficult-to-treat (D2T) rheumatoid arthritis (RA) definition in advance. This retrospective analysis included RA patients from the ATTRA registry who initiated biologic (b-) or targeted synthetic (ts-) disease-modifying anti-rheumatic drugs (DMARDs) between 2002 and 2023. Patients with D2T RA met the EULAR criteria, while controls achieved sustained remission, defined as a Simple Disease Activity Index (SDAI) < 3.3 and a Swollen Joint Count (SJC) ≤ 1, maintained across two consecutive visits 12 weeks apart. Patients were assessed at baseline and at one and two years before fulfilling the D2T RA definition. Predictive models were developed using machine learning techniques (lasso and ridge logistic regression, support vector machines, random forests, and XGBoost). Shapley additive explanation (SHAP) values were used to assess the contribution of individual variables to model predictions. Among 8,543 RA patients, 641 met the criteria for D2T RA, while 1,825 achieved remission. The machine learning models demonstrated an accuracy range of 0.606-0.747, with an area under the receiver operating characteristic curve (AUC) of 0.656-0.832 for predicting D2T RA. SHAP analysis highlighted key predictive variables, including disease activity measures (DAS28-ESR, CDAI, CRP), patient-reported outcomes (HAQ), and the duration of b/tsDMARD treatment. We identified clinical features predictive of D2T RA at baseline and up to one year before meeting the formal criteria. These findings provide valuable insights into early indicators of D2T RA progression and support the importance of earlier recognition and timely therapeutic intervention to improve long-term patient outcomes.

Department of Rheumatology 1st Faculty of Medicine Charles University Prague 128 00 Czechia

Institute of Rheumatology Na Slupi 450 4 Prague 128 00 Czechia

See more in PubMed

Fraenkel, L. et al. 2021 American college of rheumatology guideline for the treatment of rheumatoid arthritis. Arthritis Rheumatol. 73, 1108–1123 (2021). PubMed DOI

Smolen, J. S. et al. EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2022 update. Ann. Rheum. Dis. 82, 3–18 (2023). PubMed DOI

Yoshimura, Y. et al. Efficacy and safety of first-line biological DMARDs in rheumatoid arthritis patients with chronic kidney disease. Ann. Rheum. Dis. 83, 1278–1287 (2024). PubMed DOI

Kievit, W. et al. The effectiveness and medication costs of three anti-tumour necrosis factor agents in the treatment of rheumatoid arthritis from prospective clinical practice data. Ann. Rheum. Dis. 67, 1229–1234 (2007). DOI

Watanabe, R. et al. Prevalence and predictive factors of difficult-to-treat rheumatoid arthritis: the KURAMA cohort. Immunol. Med. 45, 35–44 (2022). PubMed DOI

Yoshii, I., Sawada, N. & Chijiwa, T. Clinical characteristics and variants that predict prognosis of difficult-to-treat rheumatoid arthritis. Rheumatol. Int. 42, 1947–1954 (2022). PubMed DOI

Norgeot, B. et al. Assessment of a deep learning model based on electronic health record data to forecast clinical outcomes in patients with rheumatoid arthritis. JAMA Netw. Open. 2, e190606 (2019). PubMed DOI PMC

Guan, Y. et al. Machine learning to predict Anti–Tumor necrosis factor drug responses of rheumatoid arthritis patients by integrating clinical and genetic markers. Arthritis Rheumatol. 71, 1987–1996 (2019). PubMed DOI

Phillips, P. J. et al. Four principles of explainable artificial intelligence. Sep. (2021).

Lundberg, S. M., Allen, P. G. & Lee, S-I. A Unified Approach to Interpreting Model Predictions. (2017).

Skácelová, M. et al. The beneficial effect of CsDMARDs co-medication on drug persistence of first-line TNF inhibitor in rheumatoid arthritis patients: data from Czech ATTRA registry. Rheumatol. Int. 42, 803–814 (2022). PubMed DOI PMC

World Medical Association. World medical association declaration of helsinki: ethical principles for medical research involving human subjects. JAMA 310, 2191–2194 (2013). DOI

Studenic, P. et al. American college of rheumatology/eular remission criteria for rheumatoid arthritis: 2022 revision. Ann. Rheum. Dis. 82, 74–80 (2022). PubMed DOI

Aletaha, D., Martinez-Avila, J., Kvien, T. K. & Smolen, J. S. Definition of treatment response in rheumatoid arthritis based on the simplified and the clinical disease activity index. Ann. Rheum. Dis. 71, 1190–1196 (2012). PubMed DOI

Nagy, G. et al. EULAR definition of difficult-to-treat rheumatoid arthritis. Ann. Rheum. Dis. 80, 31–35 (2021). PubMed DOI

Aletaha, D. et al. 2010 rheumatoid arthritis classification criteria: an American college of rheumatology/european league against rheumatism collaborative initiative. Arthritis Rheum. 62, 2569–2581 (2010). PubMed DOI

Chen, K., Heng, S., Long, Q. & Zhang, B. Testing biased randomization assumptions and quantifying imperfect matching and residual confounding in matched observational studies. J. Comput. Graphical Stat. 32, 528–538 (2023). DOI

van Buuren, S. & Groothuis-Oudshoorn, K. Mice: multivariate imputation by chained equations in R. J. Stat. Softw. 45, 1–67 (2011). DOI

Tibshirani, R. Regression shrinkage and selection via the Lasso. J. Roy. Stat. Soc.: Ser. B (Methodol.). 58, 267–288 (1996). DOI

Suykens, J. A. K. & Vandewalle, J. Least squares support vector machine classifiers. Neural Process. Lett. 9, 293–300 (1999). DOI

Liaw, A. & Wiener, M. The R journal: classification and regression by randomforest. R News. 2, 18–22 (2002).

Chen, T., Guestrin, C. & XGBoost: A Scalable Tree Boosting System. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. pp. 785–94. (2016).

Koo, B. S. et al. Machine learning model for identifying important clinical features for predicting remission in patients with rheumatoid arthritis treated with biologics. Arthritis Res. Ther. 23, 178 (2021). PubMed DOI PMC

Lundberg, S. M., Erion, G. G. & Lee, S-I. Consistent Individualized Feature Attribution for Tree Ensembles. ; (2018).

Kursa, M. B. & Rudnicki, W. R. Feature selection with the Boruta package. J. Stat. Softw. 36, 1–13 (2010).

Breiman, L. Random forests. Mach. Learn. 45, 5–32 (2001). DOI

Dalén, J., Svedbom, A., Hernlund, E., Olofsson, T. & Black, C. M. Identifying predictors of First-Line subcutaneous TNF-Inhibitor persistence in patients with inflammatory arthritis: A decision tree analysis by indication. Adv. Ther. 40, 4657–4674 (2023). PubMed DOI PMC

R Core Team. R: A language and environment for statistical computing [Internet]. (2023).

RStudio Team. RStudio: Integrated Development for R. [Internet]. (2019).

Wickham, H. ggplot2 [Internet]. (2009).

Wickham, H. et al. Welcome to the tidyverse. J. Open. Source Softw. 4, 1686 (2019). DOI

Henry, L. & Wickham, H. purrr: Functional Programming Tools [Internet]. (2020).

Firke, S. janitor: Simple Tools for Examining and Cleaning Dirty Data [Internet]. (2023).

Kuhn, M. Building predictive models in R using the caret package. J. Stat. Softw. 28, 1–26. (2008).

Liu, Y., Just, A. & SHAPforxgboost SHAP Plots for XGBoost [Internet]. (2023).

Chambers, S., Croft, A. P. & Singh, R. P120 Characteristics of difficult-to-treat rheumatoid arthritis: results from a systematic literature review. Rheumatology 63 (2024).

Michitsuji, T. et al. Clinical and ultrasound features of difficult-to-treat rheumatoid arthritis: A multicenter RA ultrasound cohort study. Scand. J. Rheumatol. 53, 123–129 (2024). PubMed DOI

Rashidi, H. H., Albahra, S., Robertson, S., Tran, N. K. & Hu, B. Common statistical concepts in the supervised machine learning arena. Front. Oncol. 13, 1130229 (2023). PubMed DOI PMC

Messelink, M. A. et al. Identification and prediction of difficult-to-treat rheumatoid arthritis patients in structured and unstructured routine care data: results from a hackathon. Arthritis Res. Ther. 23, 184 (2021). PubMed DOI PMC

Fujii, T. et al. Management and treatment outcomes of rheumatoid arthritis in the era of biologic and targeted synthetic therapies: evaluation of 10-year data from the KURAMA cohort. Arthritis Res. Ther. 26, 1–10 (2024). PubMed DOI PMC

Novella-Navarro, M. et al. A paradigm of difficult-to-treat rheumatoid arthritis: subtypes and early identification. Clin. Exp. Rheumatol. 41, 1114–1119 (2022). PubMed

Dey, M., Nagy, G. & Nikiphorou, E. Comorbidities and extra-articular manifestations in difficult-to-treat rheumatoid arthritis: different sides of the same coin? Rheumatology 62,1773–1779. https://doi.org/10.1093/rheumatology/keae163.161 (2023).

Rubbert-Roth, A. & Finckh, A. Treatment options in patients with rheumatoid arthritis failing initial TNF inhibitor therapy: A critical review. Arthritis Res. Ther. 11, 1–12 (2009). DOI

de Hair, M. J. H., Jacobs, J. W. G., Schoneveld, J. L. M. & van Laar, J. M. Difficult-to-treat rheumatoid arthritis: an area of unmet clinical need. Rheumatol. (Oxford). 57, 1135–1144 (2018).

Memon, S. M., Wamala, R. & Kabano, I. H. A comparison of imputation methods for categorical data. Inf. Med. Unlocked. 42, 101382 (2023). DOI

Boonstra, A. M., Preuper, H. R. S., Balk, G. A. & Stewart, R. E. Cut-off points for mild, moderate, and severe pain on the visual analogue scale for pain in patients with chronic musculoskeletal pain. Pain 155, 2545–2550 (2014). PubMed DOI