Adalimumab, an anti-TNF-α monoclonal antibody, is effective in active rheumatoid arthritis (RA), but the response is highly variable between patients. Good responders have significantly higher adalimumab serum concentrations than non-responders.1 In a recent issue of the Annals, Pouw et al2 have proposed a therapeutic cut-off concentration for adalimumab that is predictive of clinical response at a population level but, as for other TNF-α antagonists, no individualised target concentration has been defined to adjust the dose in RA.3

In a post hoc analysis we measured adalimumab concentrations in 127 samples from 30 patients with RA who received 40 mg adalimumab subcutaneously every other week.4 Concomitantly to blood sampling, disease activity score in 28 joints (DAS28) was recorded at baseline and at 6 weeks, 12 weeks, 24 weeks and 52 weeks after treatment initiation. Then, we studied the relationship between adalimumab concentration and DAS28 with a direct Emax inhibition pharmacokinetic-pharmacodynamic model, using Monolix V.3.1 (INRIA, Saclay, France). Emax models are often used to describe the concentration-response relationship, where the ‘quantity’ of response is monotonically related to drug concentration.5 With such a model, the relationship reaches an asymptote for high adalimumab concentrations when DAS28 reaches its minimal value. Estimated baseline DAS28 and adalimumab IC₅₀, that is, adalimumab concentration leading to 50% decrease of baseline DAS28, were 5.7 (coefficient of variation=10%) and 11.8 mg/L (coefficient of variation=75%), respectively. Therefore, for a patient with a median baseline DAS28=5.7, with an adalimumab concentration of 11.8 mg/L, the baseline DAS28 would be decreased twofold (figure 1). The adalimumab target concentration necessary to achieve a decrease in DAS28 from baseline to 3.2, that is, low disease activity, was obtained with the estimated parameters of the Emax inhibition model as follows: where C_target is the adalimumab concentration needed to achieve DAS28=3.2 and DAS28₀ the DAS28 at baseline. A graph expressing C_target as a function of DAS28₀ and IC₅₀ was constructed to define the individual target adalimumab concentration associated with treatment-induced low disease activity according to disease activity at baseline (figure 2). The large interindividual variability in IC₅₀ reflects the between-subject variability in sensitivity to adalimumab. Indeed, for a given adalimumab concentration, a high adalimumab IC₅₀ value results in low clinical response and vice versa. Also, given a median adalimumab IC₅₀ value, high baseline DAS28 is less associated with achieving low disease activity than moderate baseline DAS28. This finding is consistent with Aletaha et al,6 who showed that the probability of achieving remission or low disease activity decreases with increasing baseline disease activity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Relationship between disease activity score in 28 joints (DAS28) and adalimumab concentration. Each dot represents a time-point assessment of adalimumab concentration and DAS28. The open circles denote baseline (adalimumab concentrations=0), week 6, week 12 and week 24 time points. Closed circles denote week 52 time point. The solid line represents the model-predicted concentration–DAS28 relationship in an average patient. The dotted lines delineate the 5% to 95% interval.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Adalimumab concentration needed to achieve low disease activity according to baseline DAS28. This graph provides an estimation of the target concentration (C_target) on the X axis by baseline DAS28 (DAS28₀) on the Y axis, in patients with an average adalimumab IC₅₀ (solid line). The dotted lines delineate the 25% to 75% interval. For example, the C_target for a patient with a DAS28₀ of 6.45 is 12 mg/L (squares), the C_target for a patient with a DAS28₀ of 4.55 is 5 mg/L (dots). The same approach was used to achieve clinical remission, that is, DAS28<2.6 (data not shown).

The present model points out that individual disease activity at baseline is an essential factor in determining the target concentration to achieve a clinical objective. Indeed, the target concentration will be proportional to the baseline DAS28: increased in patients with very active disease and decreased in patients with moderate disease activity (figure 2). Apart from disease activity, other factors may influence the relationship between adalimumab concentration and clinical response. Hence, further studies are needed to explain the interindividual variability in IC₅₀.

We provide a tool to select an individual target concentration of adalimumab according to disease activity at baseline. This approach is hypothetical and must be replicated in a larger study before being used as a tool for therapeutic drug monitoring of adalimumab in RA.