Background Poor prognostic factors (PPF) are used as decision-criteria for treatment strategies in patients with rheumatoid arthritis (RA). However, their definition is based on the outcome of rapid radiologic progression.

Objectives To investigate the impact of PPF in RA on achieving low disease activity (LDA) at follow-up.

Methods We performed a collaborative analysis of three large German RA cohorts. Patients under routine care were either DMARD-naïve (n=991, early arthritis cohort CAPEA), on 1st conventional synthetic (cs)DMARD (n=2,547, National Database of the German Arthritis Centres (NDB)), or switching to a 2nd (n=1,959) or a 3rd DMARD (n=1,854, both biologics register RABBIT). Disease activity based on DAS28, autoantibody positivity (RF+/ACPA+), erosions, disability (HAQ≥1.2) and intake of glucocorticoids (>5mg/d) were evaluated as PPF at baseline. The outcome was DAS28 at 0, 3, 6 and 12 months. With multinomial logistic regression analyses, predictors of either LDA (DAS28<3.2) at 12 months or treatment escalation (adding or switching to cs/biologic (b)DMARD) were investigated.

Results Patients had a mean age of 57 to 60 years; 63% of DMARD-naïve patients and 71–72% of all others were female. Disease duration was 13 weeks in early RA (CAPEA) and 5–8 years in other cohorts. Patients with more treatment failures had more often PPF (not shown).

The figure shows the course of disease activity stratified by baseline DAS28, erosions and RF/ACPA. Values of DAS28 at baseline did not differ depending on presence/absence of erosions and autoantibodies. In all cohorts, irrespective of maintaining or escalating the initial treatment, patients with or without erosions and autoantibodies had similar DAS28 outcomes, whereas patients with moderate or high baseline DAS28 had higher disease activity during follow-up. The proportion of patients that did not achieve LDA at 12 months was 46% in DMARD-naïve, 34% on 1st DMARD, 47% switching to 2nd and 49% switching to 3rd DMARD.

In the multinomial model, a 1-unit increase in DAS28 was associated with a decreased probability to achieve LDA in all cohorts (table). In contrast, autoantibodies (OR from 0.8 to 1.3) and erosions (OR from 0.8 to 1.6) had no impact on achieving LDA at month 12 or on treatment escalation. The latter applied for DAS28 (OR from 0.9 to 1.5) and HAQ (OR from 0.7 to 1.2) regarding only treatment escalation.

• Download figure
• Open in new tab
• Download powerpoint

Conclusions Although autoantibodies and erosions were associated with treatment failure, their absence did not predict LDA. Instead, baseline disease activity, functional status and glucocorticoid use may be more useful to guide treatment strategies when LDA is the target.

Disclosure of Interest Y. Meissner Grant/research support from: CAPEA: funded by an unconditional grant from Pfizer. NDB: funded by unconditional grants from the German Collaborative Arthritis Centres and from a consortium of 11 pharmaceutical companies to the German Academy for Continuing Medical Education in Rheumatology. RABBIT: supported by a joint, unconditional grant from AbbVie, BMS, Celltrion, MSD, Pfizer, Roche, Samsung and UCB., K. Albrecht: None declared, A. Richter Consultant for: Pfizer, L. Baganz: None declared, D. Huscher: None declared, M. Schneider Grant/research support from: Abbvie, Chugai, UCB, Consultant for: Abbvie, Astra-Zeneca, BMS, Boehringer-Ingelheim, Chugai, Lilly, MSD, Pfizer, Roche, Sanofi-Aventis, UCB, A. Strangfeld Speakers bureau: BMS, MSD, Pfizer, Roche, Sanofi-Aventis, A. Zink Speakers bureau: AbbVie, BMS, MSD, Pfizer, Roche, UCB