Background/Purpose: The Swedish Farmacotherapy (SWEFOT) trial and other trials in
rheumatoid arthritis (RA) demonstrated that in MTX incomplete responder
patients (MTX-IR) the addition of anti-TNF was only marginally better than
stepping up to triple therapy (TT) at the group level. However, it is likely
that each treatment choice is better for some subsets of patients. We here
evaluate whether the multi-biomarker disease activity (MBDA) score could be
used to predict optimal choice of second-line treatment.

Methods: 157 MTX-IR patients from the SWEFOT trial were analyzed. Last
observation carried forward was applied in case of incomplete data. Proportions
of clinical responders at year 1 among patients with low (<30) moderate
(30-44) and high (>44) MBDA scores at randomization (Month 3) were compared
by Fisher’s exact test. Overall difference of the responses in the two therapy
arms was analyzed by Breslow-Day test (for homogeneity of odds ratio).

Results: For patients with a low MBDA (<30) score at randomization, the
likelihood of response to TT was significantly greater than to anti-TNF (88%
vs. 18%, p=0.006), whereas for patients with high MBDA (>44) the reverse was
true (35% vs. 58%, p=0.04, Figure 1A). Comparison of ROC curves with MBDA as a
predictor for each treatment identified a threshold of 38 as optimal,
corresponding to the lowest quartile of MBDA values in this population. The overall
difference of the responses to the treatments based on this cut-off (≤ vs.
>38) was highly significant (p=0.001, Figure 1B).

Conclusion: In patients with early RA and incomplete response to MTX, MBDA score
predicts the relative efficacy of subsequent treatment with triple therapy
versus anti-TNF. Perhaps most importantly, low MBDA score identifies a subset
of patients who are significantly more likely to respond to conventional triple
therapy than to anti-TNF. These findings may have major implications for the
development of more individualized and cost-effective therapeutic algorithms in
RA.