This study represents the first evaluation of the effectiveness of adalimumab in patients with spondyloarthritis (AS and PsA) who had previously been treated with one or two other TNF antagonists. The disease characteristics of the patients in both studies closely matched those of patients considered eligible for anti-TNF therapy in daily practice [31
After 12 weeks of adalimumab therapy, patients with AS or PsA having prior exposure to IFX or ETN or both experienced clinically important improvements of their diseases, as evaluated by standard outcome measures. However, response rates were generally lesser than those for patients who had not received prior anti-TNF therapy. Unadjusted and adjusted logistic regressions indicated that the likelihood of achieving BASDAI 50 and ASAS40 responses after 12 weeks was smaller for patients with than for those without a history of anti-TNF therapy. For patients with PsA, prior anti-TNF therapy also resulted in a lesser likelihood to achieve ACR50 (odds ratio, 0.65) and mPsARC (odds ratio, 0.66) responses, but this impact was not statistically important (ACR50 response rate, P = 0.130; mPsARC response rate, P = 0.194) (Table ). In the PsA study, the impact of other confounders (sex, HAQ DI, joint counts, and PsA duration) was greater than that of prior anti-TNF therapy. Fewer patients were in the PsA study than in the AS study; therefore, the likelihood of detecting statistically important differences was less. The studies were not powered to detect differences between subgroups. However, assuming a response rate of 50% in one subgroup, an odds ratio of 1.5, and sample sizes of 900 versus 300 in the subgroups, the χ 2 test would have >95% power to detect a difference in the response rates. The power would be 63% with sample sizes of 150 versus 80 and 61% with sample sizes of 350 versus 60.
Overall, the ASAS40 and BASDAI 50 response rates for patients with AS and a history of anti-TNF therapy were clinically meaningful (Table ). Notably, adalimumab therapy reduced peripheral arthritis and enthesitis; the effectiveness of adalimumab for these extraaxial manifestations was very similar for patients with and without prior TNF-antagonist treatment (Table ).
The ASAS40 and BASDAI 50 response rates were greater in patients who had been treated only with IFX therapy than in patients with only ETN therapy. However, the probability of achieving an ASAS40 response did not remarkably differ between patients with histories of treatment with IFX, ETN, or both. The likelihood of achieving a BASDAI 50 response was significantly greater for patients with prior IFX therapy.
Logistic regression analyses demonstrated that the reason for discontinuation of the prior TNF antagonist had no statistically important impact on the chances of experiencing ASAS40 or BASDAI 50 responses. However, the observed ASAS40 and BASDAI 50 response rates were greater for patients who discontinued prior anti-TNF therapy because of loss of response or intolerance than they were for those patients who discontinued the first anti-TNF agent because of lack of response. Stratifications first by type of prior TNF antagonist and then by reason for discontinuation showed similar response rates to adalimumab in patients who had lack of response to the preceding TNF antagonist. Patients who had initially responded to prior IFX and discontinued IFX because of loss of response (perhaps caused by antibodies against IFX) or because of intolerance experienced much greater BASDAI 50 and ASAS40 response rates than did patients who discontinued IFX after lack of initial response. In patients who had been treated with only ETN, stratification by reason for discontinuation showed overall similar response rates across the subsets. The interpretation of this result is limited by the small numbers of patients who had been treated only with ETN. Thus, patients who initially respond to a first TNF antagonist may be more likely also to respond to a second TNF antagonist, in particular when the type of agent is similar (monoclonal antibody). These results are generally consistent with previous analyses of patients with RA treated with adalimumab as the second or third TNF antagonist in the ReAct trial (Research in Active Rheumatoid Arthritis Trial) [5
]. ReAct, STEREO, and RHAPSODY were all designed to follow routine clinical practice for patient enrollment.
Patients with PsA who had prior IFX or ETN therapy or both achieved clinically meaningful ACR50 and mPsARC response rates of 41.7%, and 71.2%, respectively. The history of anti-TNF therapy had no relevant impact on the likelihood of achieving an ACR50 response in the unadjusted evaluations (as mentioned earlier, the adjustment for confounders did not change this result). Patients with PsA who had lack of response to prior TNF-antagonist therapy experienced somewhat less clinical improvement with adalimumab compared with patients with loss of response or intolerance to their prior TNF-antagonist treatment. Adalimumab was similarly effective for patients with prior IFX or ETN therapy. However, no logistic regression was performed owing to small group sizes. Adalimumab was also highly effective for psoriatic skin and nail lesions for patients with a history of anti-TNF therapy, with improvements similar to those of patients with PsA who were naive to TNF antagonists.
A limitation of these evaluations is the open-label study design and the explorative nature of the investigations. The response rates in patients without prior anti-TNF therapy were generally greater than the rates reported in randomized clinical trials of adalimumab [33
], ETN [35
], and IFX [37
] in AS or in PsA, in which prior treatment with other TNF antagonists was generally prohibited. For example, the ASAS40 response at week 12 was 59.3% in patients with AS without prior anti-TNF therapy in this open-label study and 39.9% in the adalimumab ATLAS trial [33
]. In patients with PsA, the ACR50 response rate was 52.3% for patients without prior anti-TNF therapy in this open-label study and 36% in the adalimumab ADEPT trial [34
]. A potential explanation for the greater response rates in our open-label, explorative studies is that the evaluations were based on observed data and omission of missing data, whereas patients with missing data were imputed to be nonresponders in the intention-to-treat analyses of randomized trial data. Furthermore, patients (and physicians) may somewhat overestimate the treatment effect in open-label studies, particularly when no alternative treatment is available, as is often the case for patients with AS.