Is there a need for new thresholds to define remission and low disease activity by Disease Activity Score 28 calculated with C reactive protein? Real life data from a local registry.

We read with great interest the recent article by Fleischmann et al which reports a post hoc analysis of five rheumatoid arthritis (RA) randomised clinical trials (RCTs) to evaluate the correlation between Disease Activity Score in 28 joints calculated by using erythrocyte sedimentation rate (DAS28-ESR) or C reactive protein (DAS28-CRP). Since its introduction, DAS28-CRP has been used to evaluate clinical response in RCTs and daily practice by applying the same DAS28-ESR thresholds to define remission and low disease activity (LDA). Consistently with some previous reports, Fleischmann et al demonstrated that DAS28-CRP underestimates disease activity when using the same remission and LDA cut-off points validated for DAS28-ESR (2.6 and 3.2, respectively). Based on these findings, the authors suggested new thresholds for DAS28-CRP (2.4 for remission and 2.9 for LDA) that significantly decrease the proportion of discordance between the two composite indices in defining remission and LDA. Still several recent RCTs employ DAS28-CRP applying the DAS28-ESR cut-off points as a primary outcome measure, and many rheumatologists use DAS28-CRP over DAS28-ESR to assess RA disease activity and implement treat-to-target strategy in daily practice. Hence, we felt it would be very useful to validate the newly introduced DAS28-CRP thresholds in a real life setting. In order to test the performance of standard versus new DAS28-CRP cut-offs, we extracted from a local registry data on all patients with RA treated with biological agents in our rheumatology unit between October 1999 and August 2013. The selected study population consisted of 562 patients receiving a total of 887 lines of biologic therapy (71 abatacept, 197 adalimumab, 15 anakinra, 45 certolizumab pegol, 230 etanercept, 18 golimumab, 237 infliximab and 74 rituximab) with a minimum follow-up period of 12 months. For each line of therapy DAS28-ESR and DAS28-CRP (using the old and the newly proposed thresholds) response rates were calculated at 6-month and 12-month checkpoints, carrying the last observation forward for patients who were missing the selected study end points. As reported in table 1, the percentage of patients achieving remission or LDA at both the checkpoints was lower for DAS28-ESR than DAS28-CRP when using standard cut-offs for both measures. As a consequence, the proportion of discordance between the two composite indices at 6-month and 12-month evaluations was 9.1% and 9.7% for remission and 13.8% and 13.7% for LDA, respectively. When clinical response rates were recalculated using the new thresholds for DAS28-CRP, the discordance between the two measures decreased (7.1% and 7.2% for remission and 9.6% and 8.2% for LDA at 6 months and 12 months, respectively) and the k coefficient of agreement increased from 0.74 to 0.79 for remission and from 0.7 to 0.77 for LDA at 6 months and from 0.75 to 0.8 for remission and from 0.71 to 0.81 for LDA at 12 months. In conclusion, our real life data confirm that the use of DAS28-ESR standard cut-offs for defining clinical response by DAS28-CRP overestimates the proportion of patients in remission or LDA, potentially affecting treatment strategies. In our daily clinical practice, the new thresholds proposed by Fleischmann et al seem to perform better, with a significant decrease of the discordance between the two indices. Additional studies are required to further validate these new DAS28-CRP cut-offs.