Table 2: Association between ACR20 response and biomarker change at Week 16.

Univariate analysis 1*Multivariate analysis
Placebo
Apremilast 20 mg BID
Apremilast 30 mg BID
Interaction
OR
(two-sided 95% CI)
valueOR
(two-sided 95% CI)
valueOR
(two-sided 95% CI)
P valueP value

TNF-α (pg/mL)0.995
(0.986, 1.003)
NS1.006
(1.001, 1.011)
0.02050.978
(0.960, 0.996)
0.01660.0024

IL-8 (pg/mL)1.000
(0.990, 1.010)
NS0.997
(0.985, 1.008)
NS0.994
(0.984, 1.004)
NSNS

IL-6 (pg/mL)0.998
(0.992, 1.004)
NS1.002
(0.991, 1.013)
NS0.989
(0.977, 1.002)
NSNS

Ferritin (ng/mL)0.996
(0.974, 1.019)
NS1.001
(0.984, 1.018)
NS0.992
(0.969, 1.016)
NSNS

vWF factor (μg/mL)1.012
(0.985, 1.040)
NS0.996
(0.938, 0.996)
0.02531.007
(0.992, 1.022)
NS0.0387

NS: not significant.
ORs, CIs, and values were calculated from a logistic regression model with percent change from baseline biomarker value at Week 16 (LOCF) as a covariate.
At Week 16 (LOCF) for the biomarkers with a significantly () different percent change from baseline in the between-treatment comparisons (apremilast 20 mg BID versus placebo or apremilast 30 mg BID versus placebo), the ACR20 (nonresponder imputation) and ACR20 (LOCF) datasets were identical.
P values were calculated from a logistic regression model with treatment as a factor, percent change from baseline biomarker value at Week 16 (LOCF) as a covariate, and interaction of treatment and percent change from baseline value at Week 16 (LOCF).