Maxime Verhoeven

51 An indirect comparison of treat-to-target strategies in early RA patients with contraindications or intolerance for MTX, TCZ as monotherapy is effective. As currently TCZ (or another bDMARD) as initial therapy in early RA is typically not possible in clinical practice, the conclusion that MTX+Pred can be regarded a similarly effective and safe treatment strategy is reassuring. A limitation of our study is the fact that this was not a head-to-head comparison and some relevant differences in baseline characteristics between trials were observed. However, all analyses were corrected for observed differences in baseline characteristics and for differences between trials by using the similar MTX arms as a reference strategy. Between the trials, there was a relevant difference in tapering protocols, with a more stringent approach to tapering in U-Act-Early. Furthermore, tapering in both trials was initiated only when being in remission over a certain time period, but the remission criteria were more strict in CAMERA-II; they were close to those of the ACR/EULAR Boolean remission criteria. 23 Tapering may have decreased disease control for the TCZ- based treatment strategies and hampered the adjustment based on reference arm. Less disease control during a later phase of the study for TCZ-based strategies can be seen in Figure 1a. Further, because VAS physician in CAMERA-II was missing, a modification of the CDAI was used which is not validated. The results based on this outcome should be interpreted with caution. Therefore, we thought it inappropriate to post-hoc switch to this measure as primary endpoint for the current study, and used the prespecified DAS28 which has been validated and is generally used in clinical trials, also in CAMERA-II and U-Act-Early. Apart from the difference of not containing an APR, m-CDAI weighs the other components differently from DAS28, which also might have influenced results. Strengths of our study are our approach of using individual patient data, enabling calculation of uniform outcomes between the two trials, the fact that the trials were performed in a similar setting and included a similar treatment strategy arm, and the appropriate use of multiple imputation and mixed models. As such, handling the clustered structure of the data and correction for important confounding is a robust method of comparing the treatment strategies and optimally using available data. Conclusion In early RA, although tight-control strategies initiating TCZ resulted in slightly better mean DAS28 and remission rates, compared to a strategy initiating MTX with 10mg prednisone, at least part of these better effects may be due to the specific effect of TCZ on APRs, as apparent from m-CDAI results. MTX+Pred and TCZ (with or without MTX) appear to have similar clinical effects as initial treatment options in tight-control strategies in early RA. 3