To examine the long-term outcome of patients with active ankylosing spondylitis (AS) clinically and by magnetic resonance imaging (MRI) after continuous treatment with the tumor necrosis factor (TNF) receptor fusion protein etanercept over 2 years.
Overall, 26 patients with active AS were treated with etanercept 25 mg twice daily subcutaneously, twice weekly with no concomitant disease-modifying antirheumatic drugs (DMARDs) or steroids. The clinical response was assessed by standardized parameters. Inflammatory spinal lesions were quantified by the ASspiMRI-a rating gadolinium-enhanced (T1-weighted gadolinium diethylenetriaminepentaacetic acid) and STIR MRI sequences. The primary outcome was a Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) improvement ≥50% after 2 years of etanercept therapy compared with the baseline value of the study.
Overall, 21 (70%) of 30 patients completed year 2. In the intent-to-treat analysis, 54% of the patients showed a 50% improvement according to the BASDAI and a 40% improvement according to the Assessment in Ankylosing Spondylitis (ASAS) criteria. In the completer analysis, 9 (43%) of 21 patients were in partial remission according to ASAS criteria. Mean ± SD BASDAI scores, which were elevated at baseline (6.3 ± 1.6), remained low: 2.7 ± 2.4 after 2 years compared with 2.6 ± 2.2 at week 54. In accordance, all other clinical parameters showed sustained improvement during year 2. The majority of patients had no disease activity flares. MRI evaluation showed a 75% improvement of active spinal lesions, but minor spinal inflammation was still present in 64% of the patients after 2 years. There were 2 serious adverse events leading to discontinuation of etanercept.
The clinical efficacy and safety of etanercept in patients with active AS without simultaneous administration of DMARDs or steroids over 2 years of continuous treatment is confirmed. Spinal inflammation as depicted by MRI decreased significantly, but a few patients still had some spinal inflammation even after long-term anti–TNF therapy.
Ankylosing spondylitis (AS), a frequent, chronic, inflammatory rheumatic disease, is the prototype and the most severe form of the spondylarthritides. AS affects young patients, most frequently starting in the third decade of life, with a prevalence of 0.1–1.1% (1). AS is characterized by spinal inflammation with sacroiliitis, spondylitis, spondylodiscitis, and spondylarthritis, but also by new bone formation with syndesmophytes and ankylosis. The socioeconomic burden of patients with AS is considerable, with a 3-fold increase in absence from work and work disability (2–4).
Therapeutic options for patients with AS have been limited during the last decades. Nonsteroidal antiinflammatory drugs (NSAIDs) are considered standard therapy. No disease-modifying antirheumatic drugs (DMARDs) are approved for the treatment of patients with AS, and there is no evidence that DMARDs are effective in axial manifestations of AS (5, 6). Sulfasalazine has some efficacy in peripheral arthritis (7, 8).
The short-term efficacy of the recombinant 75-kd tumor necrosis factor (TNF) receptor IgG1 fusion protein etanercept (Enbrel) has been demonstrated in clinical studies of patients with active AS (9–13). In some of these studies, regression of disease activity was already seen after a few weeks (9–11). The clinical efficacy was shown to last for some months to 1 year (12, 13). In correspondence, a substantial decrease of inflammatory spinal lesions as detected by magnetic resonance imaging (MRI) was detected as early as 6, 12, and 24 weeks after the initiation of therapy and persisted after 1 year of treatment (14, 15).
After the first report on successful treatment of patients with AS with etanercept over 3 months in a controlled study design (9), we have recently reported that the open readministration of etanercept after discontinuation was efficacious and safe after 1 year (13). In this report, we describe our clinical experience with the efficacy and safety of continuous therapy with etanercept over 2 years and add data of MRI examinations performed in many of these patients.
PATIENTS AND METHODS
Patients and study protocol.
After completion of the first, placebo-controlled phase of this trial (9) and experiencing a clinical relapse (13), defined as a Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) (16) score ≥4 and a score of ≥4 on a 0–10 numeric rating scale (NRS) for spinal pain, all patients (n = 26) who were known to fulfill the 1984 modified New York criteria (17) restarted treatment with etanercept continuously in the same dosage (25 mg twice daily subcutaneously, twice weekly) for a total of 102 weeks. Patients were allowed to continue to take the same or a lower dose of NSAIDs during the study. The placebo-controlled phase (9) and a 1-year followup of this cohort (13) have already been published.
Assessment of clinical response.
The disease status was assessed using internationally accepted parameters: the BASDAI, the Bath Ankylosing Spondylitis Functional Index (BASFI) (18), and the Bath Ankylosing Spondylitis Metrology Index (BASMI) (19). Physician's global assessment and patient's global assessment were also evaluated on a 0–10 NRS. Peripheral joints and entheses were clinically examined by a trained study nurse. The Short Form 36 (SF-36) questionnaire (20) was used to assess health-related quality of life. The physical and mental component summary scores of the SF-36 were calculated with the algorithm of the Medical Outcome Trust (21). The laboratory parameters C-reactive protein levels (CRP, mg/liter) and erythrocyte sedimentation rate (ESR, mm/hour) were measured by conventional means.
The efficacy of the study drug was assessed using the Assessment in Ankylosing Spondylitis (ASAS) group core set of criteria for symptomatic improvement in AS (22), which include a 20% and a 40% response according to the ASAS criteria (22) and an improvement in the 5-out-of-6 criterion (23). ASAS 40% improvement is defined as at least a 40% improvement and an absolute improvement of at least 2 units (on a scale ranging from 0 to 10) in at least 3 domains and no worsening in the fourth domain. To meet the 5-out of-6 criterion, a 20% improvement in any 5 of the following 6 domains is required: the 4 domains used for ASAS 40%, the CRP value, and spinal mobility (assessed by the BASMI score). The recently evaluated ASAS 40% and ASAS 5-out of-6 criteria were proposed to be used in the assessment of improvement in patients with AS undergoing anti–TNFα therapy because of the good discriminating capacity (23). Partial remission was defined as a score ≤2 (on a scale of 0–10) in each of the 4 ASAS 20% domains, as proposed recently (22). A status of low disease activity was defined as a BASDAI score ≤3, in accordance with a recent report by our group (24). Flares of disease activity were defined as a BASDAI score and a physician's global assessment ≥4 at one visit, similar to a recent report (24). For this evaluation, the first 2 visits (6 weeks) were excluded because they were considered to be too early for a meaningful followup to assess treatment efficacy. This resulted in an overall amount of 23 visits over 102 weeks.
The primary endpoint of the study was the percentage of patients showing at least a 50% improvement of the BASDAI at week 102 compared with baseline of the open study. Secondary endpoints were the proportion of patients who met the ASAS 40% criterion, 5-out-of-6 criterion, and criteria for partial remission as well as the mean improvement of the BASDAI, BASFI, BASMI, SF-36, CRP levels, and ESR after 102 weeks.
Magnetic resonance imaging.
MRI was performed with a 1.5 Tesla Magnetom vision (Siemens, Munich, Germany), using a spine coil and/or a body-array coil. To investigate spinal inflammation, MRI was performed as previously described (25). Sagittal views of the spine were obtained using 2 types of sequences. T1-weighted spin-echo sequences (repetition time [TR] 500 msec, echo time [TE] 14–20 msec, slice thickness 3–4 mm, 2 acquisitions) were performed before and after application of gadolinium diethylenetriaminepentaacetic acid (T1/Gd-DTPA). The evaluation of the spine was performed in 2 parts, using C2 and L5 as orientation points and always starting with the upper part. Similarly, fat-saturated STIR sequences (TR 4,000 msec, inversion time 150 msec, TE 60 msec, slice thickness 3–4 mm, 1 acquisition) were performed. Subtraction images were electronically generated as a routine procedure.
MRI examinations were performed at baseline of the initial placebo-controlled study (MR0) (9), after 6 weeks of treatment with etanercept or placebo (end of the placebo-controlled phase; MR1), after 24 weeks of continuous treatment with etanercept (open-label extension after discontinuation and retreatment; MR2), and at week 102 of the open-extension phase of the study (MR3), following a fixed protocol. MRI examinations were performed in patients from one site (Berlin) on a voluntary basis. Because not all patients agreed to undergo MRI examinations or met the scheduled appointments, the numbers of MRI examinations varied between time points. Therefore, comparisons of MRI data were performed only if pairs of MRIs of at least 2 time points were available.
MRIs were available from 15 patients at MR0, from 16 patients at MR1, from 12 patients at MR2, and from 7 patients at MR3. The patients with available MRIs did not differ from those without MRIs with respect to clinical data at baseline and response to treatment at the beginning of the study and at the time of readministration of the study drug. The MRI data have already been reported in part (14). In that report we described the 24-week followup results of the study, which had been calculated as the mean of 2 readers. In the present study, we scored all available MRIs over 2 years by 1 reader who had paired images but was unaware of the time sequence.
Scoring of MR images for active spinal lesions.
For the assessment of spinal lesions, vertebral units were evaluated by one reader (XB) using the T1/Gd-DTPA and STIR MRI sequences and were quantified with the ASspiMRI-a scoring system, as reported recently (26).
The primary analysis was performed according to the intent-to-treat principle (ITT). All patients (n = 26) who entered the open study were included in each analysis. Patients who withdrew were counted as nonresponders for every missing visit. Furthermore, a completer analysis was performed. The scoring method of Wilson was used to generate the 95% confidence intervals (95% CIs) for frequencies. For comparisons between clinical and MRI visits, the paired t-test and the paired Wilcoxon rank sum test were applied. All comparisons were 2-sided. A significance level of 5% was used. No adjustment for repeated significance testing was performed.
Of the 26 patients included at baseline, 20 patients (77%) were men. The mean ± SD age of all patients was 37.1 ± 8.9 years, with a mean disease duration of 13.7 ± 8.9 years. Twenty-three (89%) of the 26 patients were HLA–B27 positive. The mean ± SD number of swollen joints (68 swollen joints assessed) per patient was 0.7 ± 4.0 and the mean number of enthesitic regions (12 enthesitic regions assessed) per patient was 2.2 ± 2.6. Overall, 6 (23%) of 26 patients reported a positive history of uveitis (at least 1 episode). The entire 102-week extension phase was completed by 21 (81%) of the 26 patients included in the open extension at baseline (Figure 1).
An improvement in disease activity of at least 50% according to the BASDAI, as the primary outcome of the study, was reached by 53.8% (95% CI 35.5–71.2%) of the patients at week 102 (Figure 2). Similarly, 53.8% (95% CI 35.5–71.2%) of the patients were responders according to the ASAS 40% criteria, whereas 57.7% (95% CI 38.9–74.5%) of the patients were responders according to the 5-out-of-6 criteria at week 102 (Figure 2).
After having shown significant improvement at week 54 in comparison with mean values at baseline (13), the mean ± SD BASDAI values remained low during year 2 of the open extension, reaching 2.7 ± 2.4 at week 102 (P < 0.0001 compared with baseline, P = not significant [NS] compared with week 54) (Table 1, Figure 2). All single components of the BASDAI remained unchanged between week 54 and week 102 and were statistically different at week 102 in comparison with baseline (P ≤ 0.05). Spinal mobility showed sustained improvement as indicated by mean ± SD BASMI scores of 2.4 ± 2.2 (P = 0.009 compared with baseline) (Table 1, Figure 2). Also, BASFI, patient's global assessment, and physician's global assessment values remained stable on a high level compared with baseline (Table 1). The median CRP values improved from 14.0 mg/liter (range 2.0–34.0 mg/liter) to 1.0 mg/liter (range 0.0–28.0 mg/liter) at week 102 (P = 0.004 compared with baseline) (Table 1), and similarly, the median ESR values decreased from 23.0 mm/hour (range 6.0–44.0 mm/hour) at baseline to 4.0 mm/hour (range 2.0–43.0 mm/hour) at week 102 (P = 0.001) (Table 1).
Table 1. Comparison of the mean ± SD values at baseline, week 54, and week 102 of the 21 patients who completed 2 years of treatment with etanercept*
Values are the mean ± SD unless otherwise indicated. VAS = visual analog scale; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; BASFI = Bath Ankylosing Spondylitis Functional Index; BASMI = Bath Ankylosing Spondylitis Metrology Index; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate; SF-36 = Short Form 36.
Week 0 versus week 102.
Week 54 versus week 102.
Global disease assessment (0–10-cm VAS)
6.8 ± 1.8
3.0 ± 2.4
2.8 ± 2.3
6.7 ± 1.2
1.8 ± 1.6
2.0 ± 1.6
6.3 ± 1.6
2.6 ± 2.2
2.7 ± 2.4
5.8 ± 2.2
3.2 ± 2.6
3.2 ± 2.6
3.4 ± 1.9
2.5 ± 2.0
2.4 ± 2.2
CRP, median (range)
ESR, median (range)
47.1 ± 21.9
64.5 ± 26.9
70.0 ± 23.4
48.8 ± 20.4
66.9 ± 18.6
61.0 ± 20.5
After significant reduction between baseline and week 54, clinical signs and symptoms did not change during the second year of treatment with etanercept; peripheral arthritis and enthesitis were present in 1 (4.8%) of 21 patients. Anterior uveitis was observed in 1 (4.8%) of 21 patients over the entire second year of the study at week 102, compared with 2 (9.6%) of 21 patients over the first year of the study.
NSAID use was completely stopped by 7 (33%) of 21 patients during the second year of the study, which was even better in comparison with 5 (24%) of 21 patients who stopped NSAID use within the first study year. Quality of life, as assessed by the SF-36, improved significantly for all single items compared with baseline, but did not show any difference between week 54 and week 102 (Table 1).
Overall, 9 (42.9%) of 21 patients (95% CI 24.5–63.5%) were in partial remission after 102 weeks according to the ASAS remission criteria (Figure 3). Similarly, 12 (57.1%) of 21 patients (95% CI 36.5–75.5%) were in a state of low disease activity (BASDAI < 3) at week 102, which was not different from the proportion of 12 (57.1%) of 21 at week 54 (Figure 3).
There were 12 (57.1%) of 21 patients with no flares of disease activity during the study period. However, 9 (42.9%) of 21 patients showed at least 1 flare during 2 years. Five (23.8%) of 21 patients showed a flare in <5 visits (20% of all visits). The mean number of flares in the group who had flares was 8 out of 23 visits over 2 years. One patient had a BASDAI score >4 in 21 of 23 visits (see Discussion). Two of the 5 patients who discontinued had a flare when they dropped out.
Results of the MRI examinations.
Overall, inflammatory spinal lesions were detected in all patients at MR0. When using STIR, a significant improvement of active spinal lesions at MR1 was seen in the group treated with etanercept (mean ± SD ASspiMRI-a score 14.3 ± 14.1 at MR1 and 24.6 ± 20.1 at MR0, P < 0.05), but not in the placebo group (mean ± SD ASspiMRI-a score 14.1 ± 12.4 at MR0 and 13.3 ± 11.7 at MR1, P = NS). In the T1/Gd-DTPA sequence, patients in the etanercept group showed improvement of active spinal lesions, with a mean ± SD ASspiMRI-a score of 13.0 ± 10.6 at MR0 and 8.9 ± 11.9 at MR1, whereas inflammation in the patients treated with placebo did not change (mean ASspiMRI-a score 8.6 ± 10.6 at MR0 and 7.8 ± 8.7 at MR1, both P = NS).
At MR2, significant improvement of spinal inflammatory lesions was depicted by the STIR sequence in all patients irrespective of the initial treatment allocation (etanercept or placebo), with a mean ASspiMRI-a score of 5.9 ± 9.7 in comparison with 23.4 ± 18.3 at MR0 (P < 0.05). Improvement of active spinal lesions was also seen in the evaluation of the T1/Gd-DTPA sequence (mean ± SD score 7.2 ± 10.3 at MR2 versus 13.6 ± 11.5 at MR0), but the difference was not significant.
At MR3, all patients showed some improvement of active spinal lesions as compared with MR2, with a mean ASspiMRI-a score of 5.4 ± 5.1 (7.0 ± 9.2 at MR0) in the STIR sequence and 3.3 ± 5.7 (6.4 ± 9.7 at MR0) in the T1/Gd-DTPA sequence (both P = NS in comparison with MR2). In the comparison of MR0 and MR3, all patients showed improvement of spinal activity with both MRI sequences: the mean ± SD ASspiMRI-a decreased from 23.7 ± 13.3 at MR0 to 5.0 ± 5.4 at MR3 in the STIR sequence and from 11.8 ± 9.7 to 3.8 ± 6.0 in the T1/Gd-DTPA sequence (both P < 0.05) (Figure 4). However, all but 1 patient still showed at least 1 inflammatory lesion in MRI after 102 weeks. There was no correlation between the change of the MRI scores and the changes of clinical parameters throughout the study.
Serious adverse events.
Overall, there were 5 serious adverse events (SAEs) during this second year extension of the study, 2 of which led to discontinuation of the study. One adverse event was a new onset of Crohn's disease in a 28-year-old woman who started with the clinical symptom of diarrhea and was diagnosed by biopsy sample obtained at endoscopy. The patient recovered clinically after discontinuation of etanercept and treatment with corticosteroids and mesalazine. This SAE was thought to be related to etanercept. The second patient, a 56-year-old man and heavy cigarette smoker (>40 pack-years), was diagnosed with late-stage lung carcinoma after a chest radiograph had been performed because of fatigue and weight loss. The patient received further treatment in a specialized oncologic clinic. This SAE was thought to be related to smoking rather than to the treatment with etanercept. All other events were obviously not treatment related but were reported as SAEs because the patients had to be hospitalized for other reasons (Table 2).
Table 2. Drug-related adverse events and serious adverse events that occurred in 23 patients during the second year of the study*
Values are the number of patients who experienced at least 1 adverse event in each category.
Most frequent drug-related adverse events
Upper respiratory tract infection
Temporary fever of unclear origin
Infection at other sites
Infection at injection site
Elevation of liver enzyme levels
Circumscribed knots in hand/forearm region
Pruritus of the eyes
Paresthesia in forearm region
Serious adverse events
Circumscribed (nonmalignant) knots in breast region
Accident at workplace
Adverse events that were possibly drug related were reported by the study investigators to be mild to moderate. Etanercept injections were well tolerated by all patients. A temporary infection at the injection site was reported in only 1 patient. No other injection-related hypersensitivity reactions occurred (Table 2).
In this open extension study, etanercept in a dosage of 25 mg twice daily subcutaneously, twice weekly showed significant clinical efficacy in patients with active AS over 2 years of continuous treatment, confirming earlier short-term (9) and long-term results (13). The clinical results were accompanied by significant improvement of active spinal lesions as assessed by STIR and T1/Gd-DTPA MRI. This finding is consistent with those of a previous study on the same patients after only 24 weeks of etanercept treatment (14).
The efficacy was consistent because all measures of activity, function, and mobility assessed in this study remained low over 2 years, with similar values as seen after 6 and 54 weeks of treatment. Taken together, the data presented here confirm the persistent antiinflammatory effect of etanercept in patients with AS.
There was no indication of a loss of efficacy over 2 years of continuous therapy. Correspondingly, there were only a few flares: 2 of 5 patients who dropped out of the study had a flare at the time of dropout. One of those 2 discontinued due to inefficacy and the other one due to new onset of Crohn's disease. One other patient had persistently increased disease activity (BASDAI and physician's global assessment ≥4) in 21 of 23 visits. However, this patient wanted to continue treatment with etanercept because he reported considerable subjective improvement of pain and function in comparison with his situation before treatment. In comparison with baseline, his BASDAI score had improved by 20–30% at most visits, with an absolute decrease of the BASDAI of 2. This is in line with the ASAS recommendations and backs this expert proposal (22). The patient also reported a definite increase in quality of life and was able to return to work. Before anti–TNF therapy, he had had long periods of sick leave. This is in correspondence with previous reports on other anti–TNF agents (27).
Another important result of the present study is that additional treatment of patients with AS with DMARDs and steroids concomitant to etanercept treatment does not seem to be necessary. This observation is in some contrast to the results of other studies (10, 11, 28) in which DMARDs and steroids were allowed. As already mentioned, there is no evidence to support the use of steroids and DMARDs in patients with axial manifestations of AS (2, 5, 8, 29, 30).
In accordance with the clinical results, treatment with etanercept was associated with a significant reduction of active inflammatory changes of the spine as depicted by MRI. Significant regression of active inflammatory lesions was already seen after 6 weeks of etanercept therapy as compared with placebo (14). Furthermore, there was some reduction of active spinal lesions when treatment with etanercept was continued for longer periods. However, spinal inflammation was still present in some patients after 2 years of etanercept therapy. This result backs previous clinical reports (9, 13), suggesting that continuous treatment is necessary to achieve sustained suppression of clinical symptoms and spinal inflammation. The finding that in some comparisons only the STIR sequence showed significant improvement can likely be explained by the small sample size of the study (type-II error).
A lack of correlation between change of MRI scorings and clinical parameters is in line with other recent reports (31, 32). However, the overall decrease of spinal inflammatory lesions is in accordance with the improvement of clinical parameters over the same period.
Another important topic of this study was safety. Although the overall number of adverse events was not high, 2 patients had to discontinue during the second year because of SAEs. In one case, a 28-year-old woman developed a new onset of biopsy-proven Crohn's disease. This was the second case of a new onset of Crohn's disease in our study (the first case occurred after 21 weeks of therapy ), both observed in female patients, despite improvement of disease-related musculoskeletal signs and symptoms at the time of onset of Crohn's disease. Both patients recovered completely after discontinuation of etanercept and adequate treatment with corticosteroids and mesalazine. Flares and onset of intestinal bowel disease (IBD)-associated colitis have been reported in patients with spondylarthritides receiving etanercept therapy despite ongoing improvement of their axial and joint symptoms in other studies (11, 34). The recently published data suggest a low but increased risk for the development of IBD-like gut symptoms in patients with AS receiving etanercept therapy. It should be stressed that there is no evidence that these IBD cases have an especially serious or severe course. All patients responded easily to conventional therapy. Finally, etanercept, in contrast to infliximab (35), was not clinically efficacious in patients with active Crohn's disease taking conventional doses (36). The reason for this is largely unclear, although some differences between both agents have been described (37–39). This whole issue remains a challenge for the concept that the gut plays a major role in the pathogenesis of AS (40, 41). Therefore, as it stands now, the use of etanercept cannot be recommended in patients with a history or a suspicion of Crohn's disease. Discontinuation of etanercept therapy is suggested in case of a new onset of gut symptoms such as diarrhea. However, in the vast majority of patients with AS, etanercept therapy is safe and efficacious.
Regarding the second adverse event that occurred during the second year of this study, a 56-year-old man was diagnosed with advanced lung cancer. This patient was a heavy abuser of nicotine (>40 cigarettes/day for 30 years). Symptoms indicating malignancy (fatigue, weight loss, fever) occurred within a few weeks. In the literature, there is no evidence that treatment with etanercept, and with anti–TNF agents in general, is associated with malignancy (42, 43). So far carcinomas do not occur more often in patients treated with TNF blockers than in the general population (43). Therefore, the relative contribution of biologic therapy to the development of carcinoma in this patient is certainly and considerably less than that of the well-established impact of heavy cigarette smoking on this disease (44).
In summary, etanercept was efficacious in patients with active AS over 2 years of continuous treatment without concomitant DMARD and steroid therapy. There were no safety concerns. New onset of Crohn's disease may occur in patients with AS receiving etanercept treatment. Overall, anti–TNF therapy with etanercept is a major advance in the treatment of patients with AS.