ankylosing spondylitis

ARTHRITIS & RHEUMATISMVol. 58, No. 5, May 2008, pp 13241331DOI 10.1002/art.23471 2008, American College of Rheumatology

Radiographic Progression of Ankylosing Spondylitis AfterUp to Two Years of Treatment With Etanercept

D. van der Heijde,1 R. Landewe,2 S. Einstein,3 P. Ory,4 D. Vosse,2 L. Ni,5 S.-L. Lin,5

W. Tsuji,5 and J. C. Davis, Jr.6

Objective. To investigate the effect of etanercepttherapy on radiographic progression in patients withankylosing spondylitis (AS).

Methods. Patients with AS who had previouslyparticipated in a 24-week randomized, double-blind,placebo-controlled trial of etanercept therapy were en-rolled in a 72-week open-label extension. Radiographs ofthe cervical and lumbar spine from patients who re-ceived etanercept (25 mg twice weekly) for up to 96weeks were compared with radiographs from patients ina large prevalence cohort (Outcome Assessments inAnkylosing Spondylitis International Study [OASIS])who had not been treated with antitumor necrosisfactor (anti-TNF) agents. Radiographs obtained at 2time points up to 96 weeks apart from patients in bothstudy populations were digitized and read by 2 indepen-

dent readers who were blinded with regard to patientgroup and sequence. The primary end point was the96-week change in the modified Stoke AS Spine Score(mSASSS).

Results. A total of 257 patients treated withetanercept were compared with 175 unselected patientsfrom the OASIS study. There was no significant differ-ence in the change in the mSASSS from baseline amongpatients who received etanercept (mean SD 0.91 2.45) versus those from the OASIS group (0.95 3.18).

Conclusion. Unlike other inflammatory rheu-matic diseases such as rheumatoid arthritis and psori-atic arthritis, structural progression in AS seems to beindependent of TNF, despite the fact that TNF isresponsible for the signs and symptoms due to inflam-mation in this disease.

Ankylosing spondylitis (AS) belongs to a familyof rheumatic diseases known as spondylarthritides thatcharacteristically cause spinal joint inflammation andbony fusion of the spine. AS is the prototype of thespondylarthritides and is typified by ankylosis of theaxial skeleton. Radiographic damage known to resultfrom AS primarily includes fusion of entheses of thesacroiliac joints and of the posterior articulations andligaments of the spine. These fusions can lead to im-paired spinal mobility and in turn decreased ability toperform daily activities and severely reduced health-related quality of life (1).

Tumor necrosis factor (TNF) has been shownto play an important role in the inflammatory responseobserved in AS. It has been found at increased levels inthe serum and synovium of patients with AS (2,3), andtreatment with TNF-blocking agents (etanercept,adalimumab, and infliximab) has been shown to safelyand effectively reduce the signs and symptoms of AS(46) and significantly improve health-related quality oflife (1). In addition, these agents have been shown to

ClinicalTrials.gov identifier: NCT00356356.Supported by Immunex Corporation, a wholly owned subsid-

iary of Amgen Inc., and by Wyeth Pharmaceuticals. The Rheumatol-ogy Department of the Academic Hospital Maastricht received com-pensation from Amgen for the use of the OASIS data set.

1D. van der Heijde, MD, PhD: Leiden University MedicalCenter, Leiden, The Netherlands; 2R. Landewe, MD, D. Vosse, MD:University Hospital Maastricht, Maastricht, The Netherlands; 3S.Einstein, MSE: Bio-Imaging Technologies Inc., Newtown, Pennsylva-nia; 4P. Ory, MD: University of Washington, Seattle; 5L. Ni, MS, S.-L.Lin, MD, PhD, W. Tsuji, MD: Amgen Inc., Thousand Oaks, Califor-nia; 6J. C. Davis, Jr., MD, MPH: University of California, SanFrancisco.

Dr. van der Heijde has received consulting fees, speaking fees,and/or honoraria (less than $10,000 each) from Abbott, Amgen,Centocor, Wyeth, Chugai, UCB, Roche, and Schering-Plough. Dr.Landewe has received consulting fees, speaking fees, and/or honoraria(less than $10,000 each) from Abbott, Amgen, Bristol-Myers Squibb,Centocor, Wyeth, UCB, and Schering-Plough. Dr. Ory has receivedconsulting fees, speaking fees, and/or honoraria (more than $10,000 each)from Amgen, Abbott, and Targeted Genetics. Drs. Ni, Lin, and Tsuji ownstock or stock options in Amgen. Dr. Davis has received consulting fees,speaking fees, and/or honoraria from Abbott and Wyeth.

Address correspondence and reprint requests to D. van derHeijde, MD, PhD, Korte Raarberg 46, 6231 KR Meerssen, TheNetherlands. E-mail: [email protected].

Submitted for publication August 14, 2007; accepted inrevised form February 11, 2008.

1324

suppress bony inflammation as detected on magneticresonance imaging (79).

TNF also plays a significant proinflammatoryrole in rheumatoid arthritis (RA) and psoriatic arthritis(PsA), 2 inflammatory rheumatic diseases that are dom-inated by bone destruction rather than bone formation.TNF-blocking agents have been effective in reducingdisease activity as well as halting the destructive processof these diseases (1013). This relationship betweendisease activity (inflammation) and bone damage in RAis well established, and prevention of radiographic dam-age through the suppression of the inflammatory processis a widely recognized treatment goal.

A similar relationship between inflammation andbone damage has not been demonstrated in AS (14). Infact, studies in animal models of AS have suggested anuncoupling of inflammation and bone formation in thespine (15,16), and there is increasing evidence that boneformation in the spine is under the influence of bonemorphogenetic protein (16,17) and the Wnt signalingpathway (18). The role of TNF in this process is still notfully elucidated. Findings of a number of uncontrolledclinical studies have suggested that, analogous to thesituation in RA and PsA, anti-TNF agents may inhibitprogression of structural damage in AS (19,20). To date,however, these anecdotal observations are not corrobo-rated by solid evidence from controlled studies. In thepresent study, using a 3-way blinded radiograph readingdesign with 2 independent readers and adjudicationmethodology, we investigated whether 96-week radio-graphic progression in a cohort of patients treated withthe TNF-blocking agent etanercept differed from the96-week radiographic progression in an unrelated obser-vational cohort of patients who had never been treatedwith anti-TNF agents.

PATIENTS AND METHODS

Patients and study design. Patients in the etanerceptarm of this controlled study were enrolled in a 24-weekmulticenter, double-blind, placebo-controlled randomizedcontrolled trial (RCT) followed by a 72-week open-labelextension during which all patients received etanercept(ClinicalTrials.gov NCT00356356) (5). In the RCT, 277 ASpatients (139 in the placebo arm and 138 in the etanerceptarm) were followed up for 24 weeks. Eligible patients werethen given the option to enroll in an open-label extension toevaluate the safety and efficacy of etanercept treatment for upto 96 weeks. Patients were treated with etanercept 25 mgsubcutaneously twice weekly, and were allowed to receiveconcomitant nonsteroidal antiinflammatory drugs (NSAIDs),analgesics, and disease-modifying antirheumatic drugs(DMARDs) including corticosteroids. All patients from the

RCT and the open-label extension who received at least 1 doseof etanercept and had baseline radiographs of the cervical andlumbar spine were analyzed. Therefore, patients who receivedetanercept in the RCT could have received up to 96 weeks ofcontinuous etanercept treatment, and patients who receivedplacebo in the RCT could have received up to 72 weeks ofcontinuous etanercept treatment.

Patients in the control arm of the present study wereenrolled in the Outcome Assessment in Ankylosing SpondylitisInternational Study (OASIS) (21). The OASIS study was aninternational observational study on outcome in AS patientsfrom 3 different countries (Belgium, France, and The Nether-lands). It included consecutive patients who were followed upfor 10 years according to a predefined protocol, includingassessment of radiographs of the cervical and lumbar spine atbaseline, at 1-year and 2-year followup, and every 2 yearsthereafter. Patients were treated according to common prac-tice guidelines including the use of NSAIDs, analgesics, andregular exercise therapy. All patients from the OASIS studywith radiographs at baseline and at 2 years, except patientswith complete spinal fusion at baseline (n 5), were evaluatedas controls for the present study (n 175).

Procedures and end points. Radiographs of the lateralcervical and lateral lumbar spine, obtained at baseline and at96 weeks in patients in the etanercept arm and the control arm,were digitized and patient identifiers and temporal indicatorsremoved, to ensure blinding. Radiographs were scored inde-pendently by 2 trained readers who were blinded with regard totreatment, temporal sequence, and patient group, using acomputer-assisted masked reading system. Radiographs werescored using the modified Stoke AS Spine Score (mSASSS)(22), identified as the preferred radiographic scoring methodin AS by the ASsessment in Ankylosing Spondylitis Interna-tional Working Group (ASAS) and Outcome Measures inRheumatology Clinical Trials group (22,23). In the mSASSS,all anterior corners (from the lower corner of T12 to the uppercorner of S1 and from the lower corner of C2 to the uppercorner of T1) are scored for the presence of erosions, sclerosis,and/or squaring (1 point per site), nonbridging syndesmo-phytes (2 points per site), and bridging syndesmophytes (3points per site). The mSASSS, which is the sum of the scores atall individual sites, ranges from 0 to 72. In order to assessintrareader variability, 12.5% of the patients radiographs werereread by each reader. These patients were selected to equallyrepresent 4 quartiles of change from baseline, determined inthe initial reading. Intrareader reliability for status scores(assessed by intraclass correlation coefficient) was 0.9 foreach of the readers, and interreader reliability was 0.81.

The primary analysis was a direct comparison of the96-week change from baseline in the mSASSS between allpatients who had baseline radiographs in the etanercept groupand those in the control group, adjusted for baseline mSASSS.Missing information on 96-week change from baseline in themSASSS due to missing postbaseline radiographs was imputedusing the median change from baseline score observed amongcontrol patients with the same baseline score.

A secondary sensitivity analysis included a comparisonbetween all patients who had baseline and 96-week radio-graphs in the etanercept group and those in the control group.Two sensitivity analyses were performed on a subpopulation: acomparison between all patients who had baseline radiographs

RADIOGRAPHIC PROGRESSION OF AS AFTER ETANERCEPT TREATMENT FOR UP TO 96 WEEKS 1325

in the etanercept group and those from the OASIS study whowould have fulfilled the entry criteria for the RCT, and acomparison between all patients who had baseline and 96-weekradiographs in the etanercept group and those from theOASIS study who would have fulfilled the entry criteria forthe RCT.

Other sensitivity analyses involved simultaneous ad-justment for multiple characteristics of disease activity atbaseline (i.e., mSASSS, Bath Ankylosing Spondylitis Func-tional Index [BASFI] [24], Bath Ankylosing Spondylitis Dis-ease Activity Index [BASDAI] [25], and C-reactive proteinlevel) in the primary and secondary patient populations de-scribed above, or comparison of patients in the etanerceptgroup and in the control group after stratification according towhether they regularly took NSAIDs during the study. RegularNSAID use was defined as taking usual antiinflammatorydoses of medications throughout the duration of the study.Patients who discontinued regular NSAIDs, took NSAIDs onlyas needed for control of symptoms, or took NSAIDs at lessthan the usual antiinflammatory doses were not considered tobe regular NSAID users. Further comparisons included ana-lyses in which patients were stratified by duration of treatment(etanercept patients who received etanercept for 48 weeks,72 weeks, or 96 weeks), by treatment during the initial 24week RCT (patients who received placebo or etanerceptduring the RCT), and by response status (according to theASAS 40% response criteria [ASAS40]) (26).

Statistical analysis. Baseline characteristics of thepatients in the etanercept trial and patients in the OASIS studywere compared by chi-square test for categorical variables andt-test for continuous variables. Radiographic progression wascompared between the patients in the etanercept trial andthose in the OASIS study using Quade rank analysis ofcovariance, adjusted for baseline mSASSS on the change inmSASSS from baseline to 96 weeks.

RESULTS

Patient characteristics at enrollment. A total of93% of the patients (257 of 277) who had previouslyreceived either placebo (n 139) or etanercept (n 138) in the RCT enrolled in the open-label extensionand had data available for radiographic analysis. Of the257 patients, 76% received etanercept for at least 48weeks and 50% for at least 72 weeks. Figure 1 shows thedisposition of patients in the RCT and open-label exten-sion. A total of 175 of the 219 original OASIS patientshad available radiographs for analysis. Of these 175patients, none had received therapy with etanercept oranother anti-TNF agent. Forty-three percent of them(n 76) would have met the inclusion criteria for theRCT at baseline.

Demographic and baseline clinical data on theetanercept-treated patients and the control patients areshown in Table 1. Mean age, disease duration, male/female distribution, and frequency of HLAB27 positiv-

ity were similar between the 2 groups. A significantlysmaller proportion of patients from the OASIS cohortwere taking DMARDs at baseline, compared with theproportion of patients receiving etanercept (10% versus32%; P 0.0001). As would be expected, OASISpatients also had significantly less active disease, asindicated by lower values for patient global assessment,BASFI, and BASDAI (all P 0.0001). These valueswere more comparable when the analysis was limited toOASIS patients meeting the entry criteria for the RCT.Notwithstanding these differences in disease activityvariables, measures assessing severity such as themSASSS, and spinal mobility scores were similar be-tween patients in the 2 groups.

Efficacy. The changes in radiographic scoresamong patients in the etanercept arm and those in thecontrol arm (OASIS) were similar between the 2 groups(mean SD 0.91 2.45 and 0.95 3.18, respectively;P 1.00) (Table 2). The probability plot of mean change

Figure 1. Patient disposition in the initial randomized controlled trial(RCT) and the open-label extension (OLE). AE adverse event;ISR injection site reaction.

1326 VAN DER HEIJDE ET AL

in the mSASSS illustrated that the change was close to 0 inmost patients (Figure 2). When mean change in themSASSS was compared between patients who receivedetanercept and OASIS patients who met RCT entry re-quirements at baseline, they were again found to be similar,as indicated by the closely overlapping probability plots(Figure 3). In the latter group, the mean SD change was1.27 3.64 (P not significant versus patients who receivedetanercept) (Table 2). Twenty-nine patients in the etaner-cept group (11%) had missing postbaseline radiographs,while no patients in the OASIS cohort had missing post-baseline radiographs. When mean change in the mSASSSwas compared between the etanercept-treated patientswithout missing postbaseline radiographs (n 228) andthe OASIS patients, no significant difference was detected(P 0.0.83).

Changes in mean baseline cervical and lumbarradiography scores were also compared between pa-tients receiving etanercept and, as 2 separate groups,patients from the OASIS cohort overall and patientsfrom the OASIS cohort who met entry criteria for theRCT. Again, there was no significant difference in thechange in cervical or lumbar radiography scores betweenthe patients receiving etanercept and the OASIS pa-tients (P 0.28 and P 0.29, respectively). Similarly,there was no significant difference between the meanchanges in cervical or lumbar radiography scores be-tween patients receiving etanercept and the OASISpatients who met entry criteria for the RCT (P 0.66and P 0.17, respectively). The mean changes incervical or lumbar radiography scores were also found tobe similar when the OASIS patients were compared with

Table 1. Baseline characteristics of the patients*

Baseline characteristicOASIS

(n 175)

OASIS meetingRCT entry criteria

(n 76)Etanercept(n 257)

Age, mean SD years 44 12.5 48 12.3 41 10.2Male, no. (%) 121 (69.1) 54 (71.1) 194 (75.5)Duration of AS, mean SD years 11 8.5 12 9.8 10 8.5CRP, mean SD mg/dl 1.5 1.98 1.50 1.81 2.0 2.20Patients with CRP level outside normal range

of 01.00 mg/dl, no. (%)10 (5.7) 3 (3.9) 2 (0.8)

DMARD use at baseline, no. (%) 18 (10.3) 4 (5.3) 83 (32.3)Methotrexate 1 (0.6) 1 (1.3) 30 (11.7)Sulfasalazine 17 (9.7) 3 (3.9) 56 (21.8)Hydroxychloroquine 0 (0.0) 0 (0.0) 4 (1.6)

HLAB27 positive, no. (%) 142 (81.1) 64 (84.2) 201 (78.2)Patient global assessment (0100), mean SD 38 27.8 27 27.2 63 18.0BASFI, mean SD 34 25.5 55 16.6 54 20.7BASDAI, mean SD 35 20.9 47 19.8 63 20.9Modified Schober score, mean SD 2.9 1.4 2.3 1.4 3.0 1.7mSASSS, mean SD 14 17.6 19 20.8 16 18.3

* Baseline refers to the baseline of the Outcome Assessment in Ankylosing Spondylitis InternationalStudy (OASIS) or the randomized controlled trial (RCT) of etanercept; n values are the number ofpatients with baseline radiographs. AS ankylosing spondylitis; CRP C-reactive protein; DMARD disease-modifying antirheumatic drug; BASFI Bath Ankylosing Spondylitis Functional Index;BASDAI Bath Ankylosing Spondylitis Disease Activity Index; mSASSS modified Stoke AnkylosingSpondylitis Spine Score. P 0.0001 versus etanercept-treated patients.

Table 2. Changes in radiography scores at 96 weeks*

OASIS(n 175)

OASIS meetingRCT entry criteria

(n 76)Etanercept(n 257)

mSASSS 0.95 3.18 1.27 3.64 0.91 2.45Cervical radiography score 0.42 2.11 0.53 2.29 0.49 1.40Lumbar radiography score 0.53 1.88 0.73 2.00 0.42 1.84

* Values are the mean SD change from baseline. There were no statistically significant differencesbetween groups in the mean change in score. See Table 1 for definitions.


only the etanercept-treated patients who had postbase-line radiographs (as opposed to imputation of data foretanercept-treated patients from whom postbaseline ra-diographs were not available) (P 0.23 and P 0.36,respectively).

In multiple subgroup analyses, there were also nosignificant differences in the mean change in the mSASSS.When etanercept-treated patients were subgrouped ac-cording to duration of treatment, the mean change in the

mSASSS in each subgroup (48 weeks, 72 weeks, or96 weeks) was similar to the mean change in OASISpatients (P 0.98, P 0.82, and P 0.92, respectively).There was no significant difference in the mean change inmSASSS when OASIS patients were compared with eitheropen-label etanercepttreated patients who received etan-ercept during the RCT (P 0.61) or those who receivedplacebo during the RCT (P 0.55), or when OASISpatients were compared with either etanercept-treatedpatients who were ASAS40 responders (P 0.60) or thosewho were ASAS40 nonresponders (P 0.38). A total of145 of 257 etanercept-treated patients (56%) and 92 of 175OASIS patients (53%) were characterized as regularNSAID users. Stratifying patients by regular use ofNSAIDs during the study did not affect the results. Addi-tionally, adjustment for multiple covariates had no effecton the outcome.

An exploratory analysis was performed to inves-tigate the character of the radiographic changes ob-served in each cohort (Figure 4). While radiographicchanges occurred at only a small percentage of the sitesanalyzed, the prevailing type of change that did occurwas compatible with syndesmophyte formation or syn-desmophyte growth (bridging). There were no differ-ences between the 2 groups in this analysis. These resultssupport the hypothesis that progression in AS is due tosyndesmophyte formation and growth (inappropriatebone formation), rather than to erosions and squaring(bone resorption).

DISCUSSION

Results of this study, the first blinded, controlledtrial of radiographic changes in AS patients treated with

Figure 2. Probability plot of 2-year progression in the modified StokeAnkylosing Spondylitis Spine Score (mSASSS) in patients receivingetanercept and patients in the Outcome Assessment in AnkylosingSpondylitis International Study (OASIS), for whom baseline and2-year radiographs were available.

Figure 3. Probability plot of 2-year progression in the modified StokeAnkylosing Spondylitis Spine Score (mSASSS) in patients receivingetanercept and patients in the Outcome Assessment in AnkylosingSpondylitis International Study (OASIS) who met the entry criteria forthe randomized controlled trial at baseline, for whom baseline and96-week radiographs were available.

Figure 4. Distribution of the type of radiographic changes. Values arethe percentage of sites at which the given type of change (as deter-mined by change in radiography score from baseline) was noted.OASIS Outcome Assessment in Ankylosing Spondylitis Interna-tional Study.


the TNF antagonist etanercept, provide evidence thatetanercept does not inhibit syndesmophyte formation inAS. Findings of previous uncontrolled studies havesuggested that TNF blockers (both etanercept and in-fliximab) may slow the progression of structural damagein AS compared with that which occurs in patients nottaking TNF blockers (19,20). However, results of theseuncontrolled studies should be interpreted with greatcaution for several reasons. Multiple readers who wereaware of the origin of the radiographs (patient group)were used. In addition, these studies used scores thatwere assigned by readers who were blinded with regardto time order and compared them with results from theliterature that were obtained based on known timeorder. This may have a major effect on the observedprogression rate, which is substantially lower if thereader is blinded with regard to the sequence of the films(27). Finally, readers may have been biased againstprogression, knowing that all radiographs they readoriginated from trials with TNF-blocking drugs.

A randomized placebo-controlled trial wouldhave the highest validity in assessing the effect ofTNF-blocking drugs on radiographic progression. How-ever, feasibility of such a trial in patients with AS is verylimited (28). The slow rate of radiographic progressionin AS requires a minimum followup of 2 years, and anappropriate comparator with clinical efficacy for signsand symptoms similar to that obtained with TNF block-ade is not available. In addition, appropriate prognosticfactors for radiographic progression, allowing selectionof patients prone to progression, as in RA, have notbeen identified until recently (29). Also, TNF inhibitorsrapidly and effectively control clinical symptoms in alarge number of patients with AS and are now availablefor their treatment. For these reasons, a 2-year placebo-controlled study is not considered feasible or ethical bymost rheumatologists. If performed, such a study wouldbe complicated by a high dropout rate early in the trialwith many patients crossing over to TNF blockers,limiting the ability for detection of a treatment effect.The best alternative to a concurrently controlled study iscomparison with existing radiographs from patients whowere not treated with TNF-blocking drugs.

Use of a historic control group imposes specialrequirements. Historic control patients should be un-selected and representative of the entire population ofpatients with AS, should have never been treated withTNF blockers, should be well characterized, and shouldhave radiographs available with a 2-year interval byprotocol. The OASIS cohort, consisting of patients who

were enrolled in an observational study and were fol-lowed up for many years, fulfills these requirements.

Radiographs from the patients in the historiccontrol group and from the patients treated with a TNFblocker should be scored in one session. Also, readersshould be blinded with regard to the origin of theradiograph, the clinical data, and the sequence of theradiographs. Once more, all of these requirements werefulfilled in the present study.

There was a difference in disease activity betweenpatients in the OASIS study, who were unselected and hada wide range of disease activity, and patients in theetanercept group, all of whom had active disease. However,disease activity in AS may be unrelated to structuralprogression (30). In the OASIS cohort and in a separatecohort of AS patients, signs of clinical disease activity haveproven to be unrelated to structural progression as assessedradiographically (31). The validity of our observations issupported by the fact that the results were not altered whenthe comparison was limited to patients in the OASIScohort who would have fulfilled the disease activity entrycriteria for the RCT or to etanercept-treated patients whohad postbaseline radiographs (as opposed to imputing themissing data for those without postbaseline radiographs),or after adjustment for duration of exposure to etanercept,disease activity at baseline, or NSAID use.

There are methodologic issues that may have con-founded the results and deserve attention. DMARD usewas more frequent among patients in the etanercept groupthan among those in the OASIS group. There are, how-ever, no data showing that DMARDs can inhibit structuralprogression, and if so, this would result in a lower ratherthan a higher progression rate in the etanercept group. Inaddition, NSAID use was more frequent in the etanerceptgroup at baseline, and similar proportions of patients in theetanercept cohort and the OASIS cohort were classified asregular NSAID users. There has been some indication thatNSAIDs, especially if used continuously, are able to inhibitsyndesmophyte formation (32), yet as noted above, nodifference was detected when our analyses were adjustedfor NSAID use.

Based on experience regarding radiographic as-sessments of patients with RA (and PsA), many rheu-matologists expected that TNF-blocking drugs wouldinhibit structural progression in AS. These expectationsmay have been unrealistic. There are major differencesin pathophysiologic mechanisms between RA and AS:structural damage in AS is dominated by inappropriatebone formation (syndesmophytes), while in RA thedestructive process is mainly due to bone resorption.Syndesmophytes may reflect inappropriate repair that is


induced, but not necessarily maintained, by inflamma-tory stress in AS (33). In contrast to AS, the destructiveprocess that dominates in RA is well characterized andis regulated by TNF and RANKL, leading to activationof osteoclasts, while inhibitors of Wnt proteins, such asDkk-1 (18), cause a decrease in osteoblast formation.The net result is the rapid formation of erosions withoutsufficient repair. In AS, bone formation dominates thepicture. Bone formation is regulated by the transforminggrowth factor/bone morphogenetic protein family as wellas the group of Wnt proteins. Wnt signaling activatesosteoprotegerin, which counteracts RANKL-induced os-teoclast activation, a TNF-dependent process. It hasrecently been shown that Dkk-1 levels are decreased inAS and increased in RA (18), suggesting that Wntsignaling cascades are switched on in AS while beingsuppressed in RA.

These pathophysiologic considerations are sup-ported by clinical observations. In a recent investigation,as well as in this study, it was confirmed that syndesmo-phyte formation is the dominant feature of structuralprogression in AS, while erosions at the corners of thevertebrae play only a minor role (34). Moreover, we andothers have identified clear causal relationships betweeninflammation and formation of erosions in RA (35,36),while a relationship between inflammation and syndes-mophyte formation in AS could not be established. Thecurrent controlled study confirms what might be ex-pected on pathophysiologic grounds, i.e., that the for-mation of syndesmophytes may not be influenced byinhibition of TNF-regulated inflammation. Additionalevidence that TNF inhibition may not play a role insyndesmophyte formation comes from a recent prelimi-nary report that infliximab, a monoclonal antibody thatblocks TNF activity, also failed to inhibit radiographicprogression in AS patients after 2 years of use (37).

Despite all of the evidence that TNF blockadedoes not inhibit syndesmophyte formation, many inves-tigators still believe there could be a relationship be-tween inflammation as a trigger and syndesmophyteformation as a result. Such an assumption implies thatan alternative explanation for the lack of a treatmenteffect should be sought. The average disease duration ofthe etanercept-treated AS patients in this study was 10years. Assuming that inflammation triggers the initiationof syndesmophyte formation, it could be hypothesizedthat earlier intervention in the disease process, beforethe reparative processes have started, could preventfurther bone formation. Another hypothesis is thatinflammation would have to be suppressed for a longerperiod of time before the inhibitory effects can be seen

on radiographs. Such a hypothesis would gain support if,with longer followup (e.g., 4 years), a reduction inradiographic progression over time could be detected incomparison with the control group.

In conclusion, this first large, controlled study didnot demonstrate inhibition of structural progression ofspine disease in AS patients treated with etanercept fornearly 2 years, even though etanercept has been shown tobe highly effective in treating clinical signs and symptoms.This differential effect may be due to dissociation of theTNF-dependent inflammatory processes and the TNF-independent bone formation processes in AS.

AUTHOR CONTRIBUTIONS

Dr. van der Heijde had full access to all of the data in thestudy and takes responsibility for the integrity of the data and theaccuracy of the data analysis.Study design. Van der Heijde, Landewe, Einstein, Vosse, Tsuji, Davis.Acquisition of data. Landewe, Einstein, Ory, Vosse, Lin, Tsuji, Davis.Analysis and interpretation of data. Van der Heijde, Landewe,Einstein, Vosse, Ni, Lin, Tsuji, Davis.Manuscript preparation. Van der Heijde, Landewe, Vosse, Lin, Tsuji,Davis, and Dr. Marc. D. Kubasak (nonauthor; Amgen Inc.).Statistical analysis. Landewe, Ni, Lin, Tsuji.Reading of radiographs. Ory, Vosse.

ROLE OF THE STUDY SPONSORS

Marc D. Kubasak, PhD (Amgen Inc.), assisted with thewriting and preparation of the manuscript. Immunex Corporation, awholly owned subsidiary of Amgen Inc., and Wyeth Pharmaceuticalsfacilitated the study design and the writing of the manuscript, andreviewed and approved the manuscript prior to submission. Theauthors independently collected the data, interpreted the results, andhad the final decision to submit the manuscript for publication.

REFERENCES

1. Davis JC, van der Heijde D, Dougados M, Woolley JM. Reductionin health-related quality of life in patients with ankylosing spon-dylitis and improvements with etanercept therapy. ArthritisRheum 2005;53:494501.

2. Braun J, Bollow M, Neure L, Seipelt E, Seyrekbasan F, Herbst H, etal. Use of immunohistologic and in situ hybridization techniques inthe examination of sacroiliac joint biopsy specimens from patientswith ankylosing spondylitis. Arthritis Rheum 1995;38:499505.

3. Francois RJ, Neure L, Sieper J, Braun J. Immunohistologicalexamination of open sacroiliac biopsies of patients with ankylosingspondylitis: detection of tumor necrosis factor in two patientswith early disease and transforming growth factor in three moreadvanced cases. Ann Rheum Dis 2006;65:71320.

4. Braun J, Brandt J, Listing J, Zink A, Alten R, Burmester G, et al.Long-term efficacy and safety of infliximab in the treatment ofankylosing spondylitis: an open, observational, extension study of athree-month, randomized, placebo-controlled trial. ArthritisRheum 2003;48:222433.

5. Davis JC, van der Heijde D, Braun J, Dougados M, Cush J, CleggDO, et al. Sustained durability and tolerability of etanercept inankylosing spondylitis from 96 weeks. Ann Rheum Dis 2005;64:155762.

6. Van der Heijde D, Kivitz A, Schiff MH, Sieper J, Dijkmans BA,


Braun J, et al, for the ATLAS Study Group. Efficacy and safety ofadalimumab in patients with ankylosing spondylitis: results of amulticenter, randomized, double-blind, placebo-controlled trial.Arthritis Rheum 2006;54:213646.

7. Baraliakos X, Davis J, Tsuji W, Braun J. Magnetic resonance imagingexaminations of the spine in patients with ankylosing spondylitisbefore and after therapy with the tumor necrosis factor receptorfusion protein etanercept. Arthritis Rheum 2005;52:121623.

8. Braun J, Landewe R, Hermann KG, Han J, Yan S, Williamson P,et al, for the ASSERT Study Group. Major reduction in spinalinflammation in patients with ankylosing spondylitis after treat-ment with infliximab: results of a multicenter, randomized, double-blind, placebo-controlled magnetic resonance imaging study. Ar-thritis Rheum 2006;54:164652.

9. Haibel H, Rudwaleit M, Brandt HC, Grozdanovic Z, Listing J,Kupper H, et al. Adalimumab reduces spinal symptoms in activeankylosing spondylitis: clinical and magnetic resonance imagingresults of a fifty-twoweek open-label trial. Arthritis Rheum2006;54:67881.

10. Keystone EC, Kavanaugh AF, Sharp JT, Tannenbaum H, Hua Y,Teoh LS, et al. Radiographic, clinical, and functional outcomes oftreatment with adalimumab (a human antitumor necrosis factormonoclonal antibody) in patients with active rheumatoid arthritisreceiving concomitant methotrexate therapy: a randomized,placebo-controlled, 52-week trial. Arthritis Rheum 2004;50:140011.

11. Mease PJ, Kivitz AJ, Burch FX, Siegel EL, Cohen SB, Ory P, et al.Etanercept treatment of psoriatic arthritis: safety, efficacy, andeffect on disease progression. Arthritis Rheum 2004;50:226472.

12. Kavanaugh A, Antoni CE, Gladman D, Wassenberg S, Zhou B,Beutler A, et al. The Infliximab Multinational Psoriatic ArthritisControlled Trial (IMPACT): results of radiographic analyses after1 year. Ann Rheum Dis 2006;65:103843.

13. Van der Heijde D, Klareskog L, Rodriguez-Valverde V, Con-dreanu C, Bolosiu H, Melo-Gomes J, et al, for the TEMPO StudyInvestigators. Comparison of etanercept and methotrexate, aloneand combined, in the treatment of rheumatoid arthritis: two-yearclinical and radiographic results from the TEMPO study, a double-blind, randomized trial. Arthritis Rheum 2006;54:106374.

14. Marzo-Ortega H, Emery P, McGonagle D. The concept of diseasemodification in spondyloarthropathy. J Rheumatol 2002;29:15835.

15. Bardos T, Szabo Z, Czipri M, Vermes C, Tunyogi-Csapo M, UrbanRM, et al. A longitudinal study on an autoimmune murine modelof ankylosing spondylitis. Ann Rheum Dis 2005;64:9817.

16. Lories RJ, Daans M, Derese I, Matthys P, Kasran A, TylzanowskiP, et al. Noggin haploinsufficiency differentially affects tissueresponses in destructive and remodeling arthritis. Arthritis Rheum2006;54:173646.

17. Lories RJ, Derese I, Luyten FP. Modulation of bone morphoge-netic protein signaling inhibits the onset and progression ofankylosing enthesitis. J Clin Invest 2005;115:15719.

18. Diarra D, Stolina M, Polzer K, Zwerina J, Ominsky MS, Dwyer D,et al. Dickkopf-1 is a master regulator of joint remodeling. NatMed 2007;13:15663.

19. Dijkmans BA, Wanders A, van der Heijde D, Fatenejad S.Radiographic results from a long-term multicenter trial of etaner-cept (Enbrel) in patients with ankylosing spondylitis [abstract].Arthritis Rheum 2005:52 Suppl 9:S634.

20. Baraliakos X, Listing J, Rudwaleit M, Brandt J, Sieper J, Braun J.Radiographic progression in patients with ankylosing spondylitisafter 2 years of treatment with the tumour necrosis factor antibody infliximab. Ann Rheum Dis 2005;64:14626.

21. Spoorenberg A, van der Heijde D, de Klerk E, Dougados M, derVlam K, Mielants H, et al. A comparative study of the usefulnessof the Bath Ankylosing Spondylitis Functional Index and theDougados Functional Index in the assessment of ankylosingspondylitis. J Rheumatol 1999;26:9615.

22. Creemers MC, Franssen MJ, van t Hof MA, Gribnau FW, van dePutte LB, van Riel PL. Assessment of outcome in ankylosingspondylitis: an extended radiographic scoring system. Ann RheumDis 2005;64:1279.

23. Van der Heijde D, Landewe R, the ASAS Working Group. Selectionof a method for scoring radiographs for ankylosing spondylitis clinicaltrials, by the Assessment in Ankylosing Spondylitis Working Groupand OMERACT. J Rheumatol 2005;32:20489.

24. Calin A, Garrett S, Whitelock H, Kennedy LG, OHea J, MallorieP, et al. A new approach to defining functional ability in ankylos-ing spondylitis: the development of the Bath Ankylosing Spondy-litis Functional Index. J Rheumatol 1994;21:22815.

25. Garrett S, Jenkinson T, Kennedy LG, Whitelock H, Gaisford P,Calin A. A new approach to defining disease status in ankylosingspondylitis: the Bath Ankylosing Spondylitis Disease ActivityIndex. J Rheumatol 1994;21:228691.

26. Brandt J, Listing J, Sieper J, Rudwaleit M, van der Heijde D,Braun J. Development and preselection of criteria for short-termimprovement after anti-TNF therapy in ankylosing spondylitis.Ann Rheum Dis 2004;63:143844.

27. Wanders A, Landewe R, Spoorenberg A, de Vlam K, Mielants H,Dougados M, et al. Scoring of radiographic progression in ran-domized clinical trials in ankylosing spondylitis: a preference forpaired reading order. Ann Rheum Dis 2004;63:16014.

28. Van der Heijde D, Landewe R, van der Linden S. How shouldtreatment effect on spinal radiographic progression in patientswith ankylosing spondylitis be measured? Arthritis Rheum 2005;52:197985.

29. Maksymowych WP, Landewe R, Conner-Spady B, Dougados M,Mielants H, van der Tempel H, et al. Serum matrix metallopro-teinase 3 is an independent predictor of structural damage pro-gression in patients with ankylosing spondylitis. Arthritis Rheum2007;56:184653.

30. Landewe R, Wanders AJ, Mielants H, Dougados M, van der HeijdeD. Function in patients with ankylosing spondylitis is the resultant ofpatient reported disease activity and radiographic damage of thespine [abstract]. Ann Rheum Dis 2004;63 Suppl 1:889.

31. Wanders A, Landewe R, Dougados M, Mielants H, van der LindenS, van der Heijde D. Association between radiographic damage of thespine and spinal mobility for individual patients with ankylosingspondylitis: can assessment of spinal mobility be a proxy for radio-graphic evaluation? Ann Rheum Dis 2005;64:98894.

32. Wanders A, van der Heijde D, Landewe R, Behier JM, Calin A,Olivieri I, et al. Nonsteroidal antiinflammatory drugs reduceradiographic progression in patients with ankylosing spondylitis: arandomized clinical trial. Arthritis Rheum 2005;52:175665.

33. Schett G. Landewe R, van der Heijde D. Tumor necrosis factorblockers and structural remodeling in ankylosing spondylitis: whatis reality and what is fiction? Ann Rheum Dis 2007;66:70911.

34. Baraliakos X, Listing J, Rudwaleit M, Haibel H, Brandt J, SieperJ, et al. Progression of radiographic damage in patients withankylosing spondylitis: defining the central role of syndesmo-phytes. Ann Rheum Dis 2007;66:9105.

35. Boers M, Kostense PJ, Verhoeven AC, van der Linden S, for theCOBRA Trial Group. Inflammation and damage in an individualjoint predict further damage in that joint in patients with earlyrheumatoid arthritis. Arthritis Rheum 2001;44:22426.

36. Welsing PM, Landewe RB, van Riel PL, Boers M, van Gestel AM,van der Linden S, et al. The relationship between disease activityand radiologic progression in patients with rheumatoid arthritis: alongitudinal analysis. Arthritis Rheum 2004;50:208293.

37. Van der Heijde D, Landewe R, Deodar A, Baker D, Han J, Xu W,et al. Radiographic progression in patients with ankylosing spon-dylitis after 2 years of treatment not inhibited with infliximab[abstract]. Ann Rheum Dis 2007;66 Suppl II:856.


ankylosing spondylitis

Documents