The relationships between PGAxBSA and PASI and other measures of psoriasis severity, such as static PGA and DLQI, were assessed at baseline and week 16. Responsiveness to change was evaluated by Spearman correlation analyses correlating changes in PGAxBSA and PASI scores to changes in static PGA and DLQI scores.
RESULTS
Patients
The current post hoc analysis included a total of 836 patients who were randomized to receive apremilast at baseline in ESTEEM 1 (apremilast 30 mg BID: n=562) and ESTEEM 2 (apremilast 30 mg BID: n=274). Demographic and baseline disease characteristics of the included patients are summarized in Table 3.
Correlation analysis: PGAxBSA and PASI
Among patients receiving apremilast in both ESTEEM 1 and ESTEEM 2, statistically significant (P<0.0001) positive correlation coef cients were demonstrated between PGAxBSA and PASI in measuring disease severity at baseline (r≥0.757) and week 16 (r≥0.807; Table 4). Intra-class correlation coefficients values also demonstrated significant, positive agreement between PGAxBSA and PASI values among patients in both studies who received apremilast (Table 4). Both the PGAxBSA and PASI were associated with the Cohen’s effect sizes >0.8, indicating that PGAxBSA and PASI had similar measurement of treatment effect at week 16 (Table 4). Concordance of PGAxBSA and PASI Response Thresholds (improvement from baseline, 50%, 75%, 90%) At week 16 in ESTEEM 1 and ESTEEM 2, ≥79% concordance was observed between PGAxBSA and PASI (which included the total number of patients with agreement on both scales [response achievement and non-achievement]) at thresholds of 75% and 90% improvement from baseline; greater concordance (>88%) was observed using a threshold of 50% improvement from baseline (Figure 3). Most cases of non-concordance included patients who achieved the PGAxBSA response threshold, but did not achieve the same relative improvement from baseline on the PASI, indicating that PGAxBSA overrated their improvement relative to PASI (Figure 3). Similar patterns of concordance were observed for PGAxBSA and PASI using these response thresholds at week 32 in ESTEEM 1 and 2 (data not shown).
Responsiveness to Change Analyses
In ESTEEM 1 and ESTEEM 2, changes in PGAxBSA and PASI demonstrated statistically signi cant positive correlations with changes in static PGA at week 16, with all correlations of a generally similar magnitude (Table 5). Moderate positive cor- relations that were generally similar in magnitude were also observed between changes in DLQI and changes in PGAxBSA and PASI at week 16 in ESTEEM 1 and ESTEEM 2 (Table 5).
DISCUSSION
The findings from the post hoc analysis of ESTEEM 1 and 2 are consistent with and extend those of previous reports investigating the validity of the PGAxBSA as a measure of disease severity and therapeutic response.5,13,14 Initial retrospective analyses using data from the Utah Psoriasis Initiative, a registry with more than 1,200 consecutively enrolled psoriasis patients, found that PGAxBSA correlated well with PASI for the assessment of