This post hoc and exploratory analysis used data from three phase 3/4 PBO- or sulfasalazine (SSZ)-controlled RCTs of ETN (25 mg twice weekly or 50 mg once weekly, subcutaneously) for the treatment of r-axSpA, in which the primary endpoint was the proportion of patients achieving ASAS20 response. Patients received ≥ 12 weeks of ETN treatment plus optional stable background therapy (NSAIDs and/or SSZ [a disease-modifying antirheumatic drug (DMARD)], hydroxychloroquine [HCQ], or methotrexate [MTX]).

Study 16.0037 (Protocol 16.0037) was a phase 3, double-blind, 24-week trial that took place from December 2001 to October 2002, in which patients received PBO (n = 139) or ETN 25 mg (n = 138) twice weekly. The primary outcome measures were the percentages of patients achieving ASAS20 at weeks 12 and 24. Although stable doses of NSAIDs were permitted prior to and during the study as per protocol, no NSAID use was reported in this study [5]. Study 311 (NCT00421915) was a phase 3 RCT that took place from March 2002 to August 2002. Patients received ETN 25 mg (n = 45) or matched PBO (n = 39) twice weekly for 12 weeks. Evaluations were performed at weeks 2, 4, 8, and 12; 87% of patients used concomitant oral NSAIDs [6]. Study 402 (NCT00247962) was a phase 4 RCT that took place between December 2005 and February 2008. Patients received ETN 50 mg once weekly (n = 379) or SSZ titrated to a maximum of 3 g/day (n = 187) for 16 weeks. Evaluations were performed at weeks 2, 4, 8, 12, and 16; 84% of patients used concomitant oral NSAIDs. Background SSZ was not permitted [9, 13].

This analysis used data from the OLE studies of the three respective index studies: Study 16.0040 (NCT00356356; OLE of Study 16.0037), Study 312 (NCT00421980; OLE of Study 311), and Study 405 (NCT00410046; OLE of Study 402). ETN was the only study treatment in the OLE studies. The visits assessed in the OLE studies were month 6 (week 24), month 8 (week 36; week 38 for Study 405), and year 1 (week 48 only for Studies 312 and 16.0040).

This exploratory analysis assessed short-term responses by comparing the proportions of patients achieving early, intermediate, late, or no clinical response with ETN versus controls in the three index studies. Patients were categorized according to when they first achieved specific clinical improvement criteria: (1) the ASAS 5/6 criteria, (2) the ASAS40 improvement criteria, (3) improvement in ASDAS-CRP score ≥ 1.1 from baseline, and (4) improvement of Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) score ≥ 20 from baseline [14]. The ASAS 5/6 improvement criteria include six domains: patient global, pain, function (assessed by Bath Ankylosing Spondylitis Functional Index [BASFI]), inflammation (mean of BASDAI questions 5 and 6), CRP, and spinal mobility (assessed by lateral spinal flexion). To meet an ASAS 5/6 improvement, there should be an improvement of at least 20% in at least five of these six domains. The ASAS improvement criteria include four domains: patient global, pain, function (assessed by BASFI), and inflammation (mean of BASDAI questions 5 and 6). To meet an ASAS40 response, three of the four domains should improve by at least 40% and a minimum of 2 units on a scale of 0–10. In the remaining domain, there should be no worsening of ≥ 20% and a minimum of 1 unit, on a 0–10 scale. ASDAS improvement was defined as a score change of ≥ 1.1 units (equivalent to a clinically important improvement) on a 0–10 scale. BASDAI improvement was defined as a score change of ≥ 20 on a 0–100 scale [14]. An “Early” response was defined as initially occurring at week 4, and an “Intermediate” response at week 8. A “Late” response was a response initially occurring at week 12 (Studies 16.0037 and 311) or week 16 (Study 402). Patients were categorized as having a “Non-response” if they had not achieved clinical improvement by week 12 (or week 16 for Study 402). The proportions of patients achieving “Early,” “Intermediate,” “Late,” or “Non-response” for ASAS 5/6, ASAS40, ASDAS ≥ 1.1, and BASDAI ≥ 20 were compared between ETN and control arms within each study.

In the analysis of three index studies, patients were categorized according to their time to initial response (ASAS20 or ASDAS-CRP < 1.3) as “Early,” “Intermediate,” “Late,” or “Non-response” defined as described above. The time to initial response categories were subsequently stratified by concomitant DMARD use at baseline (yes vs. no) and by primary (r-axSpA) disease duration (early [< 3 years] vs. late [≥ 3 years]) because these variables are considered to have an effect on r-axSpA disease progression [15].

Long-term outcomes were analyzed by determining the proportions of patients achieving clinical response in the OLE studies based on their different temporal response (i.e., “Early,” “Intermediate,” “Late,” or “Non-response”) in the corresponding index studies. Clinical response was defined as ASAS20, ASAS40 (including ASAS20), ASDAS-CRP < 1.3, and ASDAS-CRP < 2.1 (including ASDAS-CRP < 1.3) [14]. In addition to the four index response categories, patients were reported according to their corresponding treatment sequences in index/OLE, respectively: ETN/ETN, PBO/ETN, and SSZ/ETN.

The final protocol, any amendments, and informed consent were reviewed and approved by the institutional review boards or institutional ethics committees at each of the investigational centers participating in the study. Details of the ethics committees are available upon request. All patients provided written informed consent for participation in the study. This study was conducted in compliance with the ethical principles originating in or derived from the Declaration of Helsinki and in compliance with all International Conference on Harmonisation Good Clinical Practice Guidelines.

This was an exploratory analysis based on observed cases. The analysis population was defined as all patients who had at least one dose of study drug and at least one post-baseline observation of interest. The proportions of patients achieving “Early,” “Intermediate,” “Late,” or “Non-response” in the different treatment arms within each study were compared using Cochran–Mantel–Haenszel tests without adjustment for multiplicity.

The analysis population of each OLE study was defined as all enrolled patients who received at least one dose of study drug in the OLE study and had data for at least one of the endpoints at month 6. Patients were removed from the analyses if they had an enrollment gap ≥ 12 weeks off study treatment between index and OLE studies.

Each of the four endpoints were analyzed for each OLE study: ASAS20, ASAS40, ASDAS-CRP < 1.3, and ASDAS-CRP < 2.1, reported by index response category and treatment sequence. Endpoints were analyzed at month 6, month 8, and year 1 of the OLE study stratified by four ASAS20 groups as well as by four ASDAS-CRP < 1.3 groups as defined in the index studies. Proportions of month 6, month 8, and year 1 responders and their exact 95% confidence intervals were presented using the Clopper–Pearson exact method. For each OLE study and treatment sequence, the index response categorical variable was tested using a univariate logistic regression model to determine if it predicted the month 6 response, testing the pairwise difference for each response category (“Early,” “Intermediate,” “Late”) versus the “Non-response” group per the treatment sequence, through the index and OLE studies. The same logistic regression models were used for month 8 and year 1 analyses.

The univariate logistic regression models were not adjusted for any covariates, as previous analyses showed no association between the possible covariates such as concomitant DMARDs use at baseline (Y/N) or disease duration (< 3 years vs. ≥ 3 years) and the index study response categorical variable.