fetching data ... 
Background: Predicting the prognosis after biological or targeted synthetic disease-modifying antirheumatic drug (b/tsDMARD) administration is essential in treating patients with rheumatoid arthritis (RA). However, we know little about determinant factors at the initiation of b/tsDMARD.
Objectives: This study aims to find key factors at the initiation of b/tsDMARD in predicting responses after administration.
Methods: The study recruited outpatients with RA to whom b/tsDMARD was administered from July 2014 to June 2023 and consecutively continued for at least one year. A 28-joint disease activity score using C-reactive protein (DAS28) was used for clinical evaluation as a disease activity indicator. Patients’ demographic background, such as sex, age at initiation (baseline; BL), disease duration of RA (DD), concomitant usage and dosage of methotrexate (MTX), mean dosage of glucocorticoid (GC), Sharp/van der Heijde score (SHS), the action of b/tsDMARD, order number of the b/tsDMARD, patient-related outcomes, such as Health Assessment Questionnaire Disability Index (HAQ) and pain score using a visual analog scale (PS-VAS) were collected. Clinical blood test results, such as titer and positivity of anti-citrullinated polypeptide antibodies (tACPA and pACPA), titer of rheumatoid factor (RF), C-reactive protein (CRP), matrix metalloprotease-3 (MMP-3), blood hemoglobin level (HGB), platelet count (PLT), lymphocyte count (Lymph), serum albumin level (ALB), and liver dysfunction score represented with fibrosis four index (FIB4) were also collected. They were monitored every three months until 12 months, and the association between the improvement of DAS28 at each time point from the BL and these parameters at the BL was evaluated every three months using a linear regression analysis. The association between an improvement of DAS28 from the BL to month 3 was assessed for each action of b/tsDMARDs. Statistically significant parameters using a univariate model were picked up and then examined in the significant parameters using a multivariate model. The correlation coefficients of each parameter and the parameter’s value were multiplicated, and no statistical significance was demonstrated. The values of each parameter were summed. This value was determined as each patient’s predicted value (PV). The relationships between the improvement of DAS28 at each time point and the PV for each b/tsDMARD were evaluated using a univariate model linear regression analysis. A receiver operation characteristic analysis (ROC) was conducted to fulfill the EULAR response with “good” or “moderate” (rEULAR) at month 3 regarding the PV. Chi-square tests (CST) were conducted regarding the fulfillment of the rEULAR at each time point and the cut-off index (COI) of the PV determined by the ROC to assess sensitivity and specificity. CST was performed to compare. Statistical significance was set at less than 5%.
Results: A total of 373 patients, 162 tumor necrotizing factor inhibitors (TNFi), 55 interleukin-6 inhibitors (IL6i), 51 cytotoxic T-lymphocyte antigen-4 Immunoglobulin (CTLA4-Ig), and 105 Janus kinase inhibitors (JAKi) were recruited. Of these, 83 were male (22.3%), and 290 were female (77.7%). The mean age at the BL was 64.3 years old in the TNFi, 67.1 in the IL6i, 69.5 in the CTLA4-Ig, and 70.9 in the JAKi. There were no significant differences in the demographic background, including MTX administration ratio and mean dosage among the groups. A linear regression analysis with a univariate model revealed that significant parameters in the TNFi were the dosage of MTX, the dosage of GC, MMP-3, HGB (negative correlation), PLT, ALB (negative correlation), and FIB4 (negative correlation), and in these HGB was the only significant parameter using a multivariate model. In the IL6i, DD, the dosage of MTX, SHS (negative correlation), CRP, PLT, and Lymph (negative correlation) were the significant parameters using a univariate model. The MTX dosage and CRP were the significant parameters using a multivariate model. In the CTLA4-Ig, DD (negative correlation), pACPA, administration ratio of MTX, PS-VAS, and FIB4 were the significant parameters using a univariate model, and all these parameters except for the DD were significant using a multivariate model. In the JAKi, the order number of the b/tsDMARD (negative correlation), DD, administration ratio of MTX, the dosage of MTX (negative correlation), CRP, PS-VAS, HGB (negative correlation), PLT, ALB (negative correlation), and FIB4 (negative correlation) were the significant parameters using a univariate model. PS-VAS was the only parameter with statistical significance using a multivariate model. With these parameters, the regression coefficients were 0.519 in the TNFi, 0.813 in the IL6i, 0.665 in the CTLA4-Ig, and 0.643 in the JAKi. When using the PV, the regression coefficients were 0.522 in the TNFi, 0.691 in the IL6i, 0.414 in the CTLA4-Ig, and 0.569 in the JAKi. The ROC results showed the COI and the area under the curve (AUC) were -0.904 and 0.744 in the TNFi, -1.286 and 0.816 in the IL6i, 1.442 and 0.767 in the CTLA4-Ig, and -0.38 and 0.726 in the JAKi. In the CST, the sensitivity and specificity were 85.1% and 47.1%, 86.6% and 40.0%, 85.1% and 38.6%, and 86.6% and 40.0% in the TNFi, 80.0% and 83.3%, 85.7% and 75.0%, 88.6% and 66.7%, and 85.7% and 75.0% in the IL6i, 90.0% and 47.6%, 93.3% and 42.9%, 96.7% and 57.1%, and 93.3% and 42.9% in the CTLA4-Ig, and 80.0% and 60.0%, 88.3% and 52.5%, 86.7% and 60.0%, and 85.0% and 57.5% in the JAKi, at month 3, 6, 9, and 12, respectively.
Conclusion: A novel approach to predicting the outcome of b/tsDMARD is available. This approach’s advantage is that it can be used up to 12 months after initiation. We could predict which action is most appropriate if we use these indices at the initiation of b/tsDMARD.
REFERENCES: NIL.
Acknowledgements: NIL.
Disclosure of Interests: None declared.
© The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license (