Abstract
Objective Patients with axial spondyloarthritis (axSpA) in clinical remission tapered tumor necrosis factor inhibitor (TNFi) therapy according to a clinical guideline. Over a 2-year follow-up period, we aimed to investigate flare frequency, dose at which flare occurred, type of flare, and predictors thereof.
Methods Patients in clinical remission (Bath Ankylosing Spondylitis Disease Activity Index [BASDAI] < 40, physician global score < 40, and without disease activity the previous year) tapered TNFi to two-thirds the standard dose at baseline, half at week 16, one-third at week 32, and discontinued at week 48. Flares were defined as BASDAI flare (BASDAI ≥ 40 and change ≥ 20 since inclusion), and/or clinical flare (development of inflammatory back pain, musculoskeletal or extraarticular manifestations, and/or Ankylosing Spondylitis Disease Activity Score [ASDAS] ≥ 0.9), and/or magnetic resonance imaging (MRI) flare (≥ 2 new or worsened inflammatory lesions).
Results Of 108 patients, 106 (99%) flared before 2-year follow-up: 29 patients (27%) at two-thirds standard dose, 21 (20%) at half dose, 29 (27%) at one-third dose, and 27 (25%) after discontinuation. Regarding type of flare, 105 (99%) had clinical flares, 25 (24%) had BASDAI flares, and 23 (29% of patients with MRI at flare available) had MRI flares. Forty-one patients (41%) fulfilled the Assessment of SpondyloArthritis international Society (ASAS) definition of clinically important worsening (≥ 0.9 increase since baseline). Higher baseline physician global score was an independent predictor of flare after tapering to two-thirds (OR 1.19, 95% CI 1.04-1.41, P = 0.01). Changes in clinical and/or imaging variables in the 16 weeks prior to tapering did not predict flare.
Conclusion Almost all (99%) patients with axSpA in clinical remission experienced flare during tapering to discontinuation, but in over half of these patients, flare did not occur before receiving one-third dose or less. Higher physician global score was an independent predictor of flare.
- axial spondyloarthritis
- drug tapering
- flare
- magnetic resonance imaging
- patient-reported outcomes measures
- tumor necrosis factor inhibitors
Axial spondyloarthritis (axSpA), which includes ankylosing spondylitis (AS) and nonradiographic axSpA (nr-axSpA), is a chronic inflammatory disease affecting the sacroiliac joints (SIJs) and spine, causing pain, fatigue, and stiffness.1 The primary goal of treating patients with axSpA with biologic disease-modifying antirheumatic drugs (bDMARDs) is to maximize long-term health-related quality of life through control of symptoms and inflammation and to maintain function.2,3 Previous tapering studies have reported that some patients can maintain inactive or low disease activity on a lower dose than standard dose.4-7 In Denmark, bDMARD costs are borne by public hospitals and all patients with axSpA receiving bDMARDs are mandatorily registered in the Danish Rheumatology Quality Registry, DANBIO.8 In the Capital Region, a guideline for mandatory tapering of bDMARDs for patients with axSpA in stable remission (Dose Adjustment of Biological Treatment in SpA Patients [DOBIS]) was implemented. We have previously shown that almost all patients with axSpA in clinical remission in routine care, who had their bDMARD systematically tapered until discontinuation, flared, but approximately half of them were on a lower dose than standard dose at 2-year follow-up.9 To avoid unnecessary dose tapering attempts, it would be desirable to be able to identify patients who would flare after a tapering attempt at a certain dose.
In the current report from the abovementioned prospective observational quality control study, we aimed to (1) study the frequency of flare at each dose step; (2) study types and subtypes of flare (ie, defined clinically, by Bath Ankylosing Spondylitis Disease Activity Index [BASDAI] or magnetic resonance imaging [MRI]) and changes in clinical and MRI variables from baseline to flare and from flare to year 2; (3) investigate predictors of flare within 16 weeks after reduction to two-thirds, one-half, one-third, and no dose; and (4) investigate if changes occurring during the previous tapering period predicted flare.
METHODS
The DOBIS guideline was implemented in the Capital Region of Denmark between June 2017 and July 2018. Patients were included if they had (1) initiated TNFi due to a diagnosis of axSpA, including AS or nr-axSpA, as registered in DANBIO; (2) received biological therapy in stable standard dose or lower dose ≥ 1 year; (3) had no or low clinical disease activity, defined as BASDAI < 40 and visual analog scale (VAS) for physician global score < 40; (4) had no evidence, as listed in DANBIO, for moderate or high disease activity during the past year; and (5) were judged to be in remission by their treating rheumatologist. Patients were not included if they (1) were treated with oral, intraarticular, intravenous, or intramuscular corticosteroid within 3 months, unless given orally in a stable dosage ≤ 7.5 mg daily due to nonrheumatological conditions; or (2) had major illness or conditions that independently of axSpA would require treatment with a biological drug, or as judged by the treating rheumatologist would be markedly worsened, if the patient followed the guideline.
TNFi was reduced at scheduled visits to two-thirds of standard dose at baseline, to half of standard dose at week 16, to one-third of standard dose at week 32, and was discontinued at week 48 (Supplementary Table S1, available with the online version of this article). Patients who experienced a flare stopped tapering and were escalated to the previous dose. If a patient did not regain remission, the dose was escalated to the next step in the algorithm every 4 months until the patient regained remission. If patients experienced a flare, they were scheduled for additional visits at weeks 16 and 32. MRIs were obtained at baseline, week 16, week 48, week 96, and at flare, and conventional radiographs were obtained at baseline and week 96. The SIJs and spine were visualized with semicoronal T1-weighted turbo spin-echo images and short-tau inversion recovery (STIR) images of the SIJs and sagittal T1-weighted turbo spin-echo images and STIR images of the spine. Anterior-posterior radiographs of SIJs were acquired with the patient supine, and lateral radiographs of the lumbar (Th9-S2) and cervical (C1-Th9/Th10) spine were acquired with the patient standing. A flare was defined as either a BASDAI flare (an absolute BASDAI ≥ 40 and change ≥ 20 since inclusion), a clinical flare (signs of inflammatory back pain, development of musculoskeletal manifestations such as peripheral arthritis, enthesitis or dactylitis, and/or development of extraarticular manifestations such as anterior uveitis, psoriasis, or inflammatory bowel disease, and/or an increase in ASDAS ≥ 0.9), or an MRI flare according to a predefined MRI flare definition (Supplementary Table S2 and Table S3). The MRIs were evaluated according to the predefined flare definition within 1 week by 1 of 6 preselected, experienced musculoskeletal radiologists (trained in the definition) and the assessments (presence vs absence of MRI flare) were available to the rheumatologist at the clinical visit.
The following patient-reported outcomes (PROs) were assessed at each visit: BASDAI,10 Bath Ankylosing Spondylitis Functional Index (BASFI), patient global VAS score, patient pain VAS score, patient fatigue VAS score, Health Assessment Questionnaire (HAQ),11 HAQ for the Spondyloarthropathies (HAQ-S),12 EuroQol-5 Dimension questionnaire (EQ-5D),13 and Assessment of SpondyloArthritis international Society Health Index (ASAS HI).14 ASDAS was also registered at each visit.
For the current paper, all MRIs and radiographs were read by experienced readers in known time order, but readers were blinded to clinical and other imaging data. One experienced reader evaluated the MRI SIJs according to the Spondyloarthritis Research Consortium of Canada (SPARCC) SIJ Inflammation Index15 and SPARCC SIJ Structural Score (SSS)16 for fat, erosion, backfill, and ankylosis. The SPARCC SSS for erosion, backfill, and ankylosis were used to calculate axSpA MRI SIJ Composite Structural Damage Scores.17 One experienced reader evaluated the spine according to SPARCC MRI index for assessment of spinal inflammation18 and the Canada-Denmark MRI scoring system of the spine in patients with axSpA (CANDEN).19-24 Radiographs were assessed according to modified New York (mNY) criteria25 and modified Stoke Ankylosing Spondylitis Spine Score (mSASSS)26 by 1 experienced reader.
Statistical analyses. The entire study population, including subgroups (eg, patients with vs without flare), were analyzed. Descriptive statistics were reported as n (%) for categorical variables and as mean (SD) or median (IQR) for continuous variables. Differences between patients with flare, stratified by the dose on which they experienced a flare, were compared with the chi-square test or Fisher exact test for categorical variables and Kruskal-Wallis test for continuous variables. Changes in clinical and imaging variables from baseline to flare visit and from flare visit to 2-year visit were tested by Wilcoxon signed-rank test or binomial sign test, as appropriate. P < 0.05 was considered statistically significant.
Potential predictors of the independent variable (ie, flare, including all flare types) were investigated by logistic regression models in defined prediction cohorts (Supplementary Figure S1, available with the online version of this article). Univariable and multivariable analyses were done to predict flare within 16 weeks after a tapering attempt to a certain dose (ie, two-thirds, half, and one-third dose). Only patients with complete baseline MRI or baseline MRI missing at random (n = 100) were included in the prediction analyses; patients with MRI not done due to hyperkyphosis were not included. Missing values were imputed by multiple imputation by chained equations (20 imputed datasets). Fourteen variables considered clinically relevant by expert opinion among investigators were assessed as independent variables. Sex, smoking (current vs previous or never smokers), HLA-B27 positivity, and mNY criteria positivity were included as binary independent variables, whereas age, time since diagnosis, number of previous bDMARDs, patient pain, physician global VAS, ASDAS, SPARCC SIJ Inflammation Index, CANDEN total inflammation, SPARCC SSS erosion, and CANDEN fat were included as continuous ones. Variables with a P value < 0.10 in univariable analyses were included in the initial multivariable model where backward selection was performed in stacked imputed datasets with weighted regression using the likelihood ratio test.27 In order to assess the potential significance of baseline variables excluded in univariable analyses, these variables (1 at a time) were reintroduced into the multivariable model. The area under the receiver-operating characteristic curve was estimated by internal validation. A significance level of P < 0.05 was applied in the variable selection procedure. The results of logistic analyses were presented by odds ratio (OR), 95% CI of the OR, and P value of the likelihood ratio test. Analyses were performed by a statistician in R software version 4.1.0 (R Foundation for Statistical Computing).
Statement of ethics and consent. The intervention was done based on a clinical guideline and therefore it did not require ethical committee approval or signed informed consent from patients. The Danish Clinical Quality Program – National Clinical Registries (RKKP) approved the collection of quality data from DANBIO (DANBIO-2019-07-12) and the Danish Data Protection Agency approved the collection and handling of data (RH-2017-216, I-Suite no. 05656).
RESULTS
Occurrence of flare. Of the 109 patients that started tapering according to the DOBIS guideline,9 108 patients were investigated in the current study (1 patient missed some tapering steps due to pregnancy and was not included; Supplementary Figure S2, available with the online version of this article). Baseline descriptive statistics of all patients and the patients receiving standard doses have been previously described.9 Of the 106 patients who experienced a flare, 29 patients (27%) flared at two-thirds of standard dose, 21 (20%) at half dose, 29 (27%) at one-third dose, and 27 (25%) after discontinuation. One patient completed the 2-year follow-up period without a flare, and 1 patient died of multiorgan failure due to infection and comorbidity before week 32 without having experienced a flare. Besides this, flares in axSpA were the only adverse events registered. Comparison of baseline characteristics for patient groups, stratified by the dose on which the patients flared, is shown in Table 1.
Baseline characteristics of patients who experienced a flare, stratified by the dose that they experienced flare on (n = 106).
Description of flare. Of 106 patients experiencing flare before the 2-years follow-up, 105 (99%) patients had a clinical flare, 25 (24%) had a BASDAI flare, and 23 (29% of patients with MRI at flare) had an MRI flare (Figure 1; Supplementary Table S4, available with the online version of this article). Forty-one patients (41% of patients with available data) experienced an ASDAS flare (≥ 0.9 increase in ASDAS since baseline) and the most frequent symptoms were back and buttock pain (n = 93, 89%) and pain in peripheral joints or entheseal regions (n = 48, 46%; Supplementary Table S4).
Venn diagram showing flare types (n = 106). The left schematic shows the overlap between the 3 flare types (ie, BASDAI, clinical, and MRI flare). The right schematic shows overlap between BASDAI flare, MRI flare, and clinically important worsening according to ASDAS. BASDAI flare is defined as a change in BASDAI ≥ 20 since inclusion and an absolute BASDAI ≥ 40. Clinical flare is defined as signs of inflammatory back pain, development of musculoskeletal manifestations such as peripheral arthritis, enthesitis or dactylitis, and/or development of extraarticular manifestations such as anterior uveitis, psoriasis, or inflammatory bowel disease, and/or an increase in ASDAS ≥ 0.9. MRI flare is defined as ≥ 2 new or worsened inflammatory lesions in the sacroiliac joints or spine according to a predefined MRI flare definition. ASDAS: Ankylosing Spondylitis Disease Activity Score; BASDAI: Bath Ankylosing Spondylitis Disease Activity Index; MRI: magnetic resonance imaging.
Comparison of variables at baseline and at flare showed that all variables used for assessing disease activity were statistically significantly worsened at flare compared to baseline and improved at 2-year follow-up (Table 2). Additionally, all PROs were worsened at time of flare compared to baseline and statistically significant improved at 2-year follow-up compared to time of flare. Almost all the inflammatory MRI variables had increased statistically significantly at flare, whereas all structural MRI scores, except the CANDEN spine fat score, did not change. Imaging variables that were statistically significantly higher at flare compared to baseline included SPARCC spine total inflammation index, CANDEN MRI spine inflammation index, and CANDEN MRI spine fat score (Table 2). As mentioned, 23 patients fulfilled the predefined MRI flare criterion. In addition to this, 13 patients worsened in MRI spine inflammation (12 patients) and/or MRI SIJ inflammation (1 patient; Table 3). In total, 40% of all patients with an MRI performed at time of flare showed an increase in MRI spine and/or SIJ inflammation score (Table 3). Of the patients with ASDAS flare, 52% showed an increase in MRI spine and/or SIJ inflammation score (Table 3) at time of flare.
Values at flare vs values at baseline and at 2-year follow-up.
Increases in various MRI inflammation scores from baseline to time of flare, in all patients, in patients with ASDAS increase ≥ 0.9, and in patients without MRI flare.
Flare occurred on average after 218 (SD 118) days, and the number of days from last dose reduction to flare was a mean of 99.1 (SD 44.3) and median of 106 (IQR 70-120). An MRI example of flare in the spine and in the SIJ is shown in Figure 2.
T1-weighted (left) and STIR (right) MR images of the spine and SIJs at baseline (upper row) and at flare (lower row). At flare, inflammatory lesions (arrows) are visible in the spine and at sacral and iliac sides in both SIJs. MR: magnetic resonance; SIJ: sacroiliac joint; STIR: short-tau inversion recovery.
Prediction of flare. Univariable and multivariable regression analyses including potential predictors at ‘baseline’ of a dose tapering were done in predefined groups (Supplementary Figure S1, available with the online version of this article). In these analyses, higher physician score was an independent predictor for flare after tapering to two-thirds of standard dose (OR 1.19, 95% CI 1.04-1.41, P = 0.01; Table 4). When assessing flare after tapering to half and one-third dose, the prediction analyses were unstable (data not shown).
Univariate and multivariable logistic regression analyses for (A) prediction of flare within 16 weeks after tapering to two-thirds dose and for (B) predictive value of changes within 16 weeks prior to tapering to half dose.
Changes within 16 weeks after the previous dose reduction were investigated as continuous predictors of flare after the current dose reduction; however, no statistically significant predictors were found from baseline to week 16 (Table 4). At subsequent dose steps, samples were not sufficient to investigate potential predictors.
DISCUSSION
Our study describes patients with axSpA in clinical remission who tapered TNFi according to a clinical guideline. Flare frequency, the TNFi dose at which flare occurred, and flare characteristics were investigated, as were potential clinical and imaging predictors of flare. At the 2-years follow-up, almost all patients had experienced a disease flare, but approximately half of the patients did not experience a flare until tapering to one-third dose. Higher physician global score was the only independent predictor of flare within 16 weeks after the first tapering step (to two-thirds of standard dose).
Within the 2-year follow-up period, 106 (99%) patients experienced a flare, of which approximately one-fourth flared at two-thirds of the standard dose, one-fifth at half dose, one-fourth at one-third dose, and one-fourth after discontinuation. No previous studies have described stepwise dose tapering until withdrawal according to a clinical guideline, except that a previous report from the current quality control study showed that approximately half of the patients received lower dose than standard dose at 2-years’ follow-up.9 Previous tapering studies reported flare in the majority of patients when discontinuing bDMARDs,28-33 which is in accordance with our study. Other studies reported flares in 58-84% of patients when tapering the dose to two-thirds,6,34-36 which is overall comparable to our study where 73% did well on two-thirds of standard dose. In 2 randomized controlled trials (RCTs), 79-83% of patients did well on a half dose after 40-48 weeks,37,38 which is more than in our study, where 53% did well on half of the standard dose.
In the current study, disease flare was defined in 3 different ways (ie, patient-reported BASDAI flare defined as BASDAI ≥ 40 and increase ≥ 20 since baseline, MRI flare based on a consensus-based flare definition, and as a clinical flare including certain symptoms and/or an ASDAS increase of ≥ 0.9 since baseline [ie, ASAS definition for clinical important worsening]). Previous tapering studies included BASDAI-based flare definitions similar to ours.34,39 To our knowledge, there is no validated MRI flare definition and no previous study has used an imaging-based flare definition. The threshold for MRI worsening in our study was deliberately set high to avoid the possibility that patients stopped tapering due to minor, potentially unimportant MRI worsening. It should be noted that 40% of all patients experienced an increase in inflammation in the spine and/or SIJs, whereas only 29% fulfilled the predefined MRI flare criteria, suggesting that the applied MRI criteria may have been too strict. At flare, the spinal inflammation scores increased in some patients, whereas SIJ inflammation increased in others, which indicates that an MRI for assessing the inflammation status of a patient with axSpA should include both the spine and SIJs. To our knowledge, only 1 study has included the ASAS-defined cut-off for clinically important worsening (ASDAS ≥ 0.9) in their flare definition,40 encouraging us to investigate this cut-off. This open-label, 1-year follow-up study randomized patients with psoriatic arthritis and axSpA with stable low disease activity (LDA) on TNFi to a treat-to-target strategy with or without tapering, with LDA at 1 year (measured by ASDAS in patients with axSpA) as the primary endpoint.40 The proportion of patients with LDA at 12 months was noninferior in the tapering group, despite lower average TNFi dose and approximately 25% of patients having discontinued TNFi. An RCT38 included clinically important worsening in posthoc analyses but did not include ASDAS ≥ 0.9 in the prespecified flare definition. In the current study, 41 patients (41%) had clinically important worsening according to ASDAS at the time of flare. Some patients with clinical flare had an MRI flare exceeding the predefined criteria without having an ASDAS flare. This suggests that the ASDAS flare definition does not identify all patients experiencing flare when used alone, and that using both MRI and ASDAS for objectively identifying flare may be appropriate.
Clinical, laboratory, MRI, and patient-reported variables for disease activity were statistically significantly worsened at flare compared to baseline. Correspondingly, almost all these variables were statistically significantly improved at 2-year follow-up compared to the time of flare, in agreement with the increased therapy. Previous studies also reported improved disease activity measures as well as improved PROs at follow-up in patients after dose escalation or reinitiation of biologic therapy. However, no studies included all the different PROs investigated in our study. Regarding MRI variables, no tapering studies have reported MRI variables at flare.
In our study, the time from last dose reduction to flare was about 14 weeks. Previous studies have reported that patients flared at approximately 16 weeks after tapering,29,41,42 which is only slightly longer than reported in the current study. Our patients were followed every 16 weeks and were tapered further if no flare had occurred. Thus, we are not able to tell whether the different variables for disease activity, including ASDAS, would have escalated further, potentially leading to a flare, if patients had remained on the same bDMARD dose.
Only a few studies reported results of multivariable regression analyses in patients with axSpA tapering bDMARDs. In an RCT, posthoc analyses identified HLA-B27 negativity as a predictor of flare in patients who tapered their TNFi, but not in patients who withdrew therapy.42 In a retrospective study investigating loss of remission and disease flares in patients with axSpA receiving TNFi, residual disease activity was associated with disease flare.43 In the present study, physician global score at baseline was an independent predictor of flare within 16 weeks after tapering to two-thirds dose. Our finding suggests that the treating physician is frequently able to interpret the patients’ overall status. However, the physician’s evaluation may potentially also have influenced whether a patient was later registered as having a flare or not (ie, caused bias). It would have been interesting to investigate which different features the rheumatologist considered when assessing the physician global score.
It may be possible that changes occurring after the previous dose reduction (eg, a subclinical worsening of a clinical or MRI outcome) could indicate or predict whether further dose reduction would cause flare. Consequently, we investigated if changes within 16 weeks prior to dose reduction were predictors of flare. However, none of these changes were predictive. No previous studies have reported comparable investigations.
Strengths of this study include the fact that patients originated from a routine care setting, with less strict inclusion and exclusion criteria than in RCTs, and data are therefore expected to be more generalizable to real-world patients treated in routine care. Further, patients were studied with a highly systematic approach with a standardized tapering regime and flare definition. Limitations include the relatively small sample size. Further, since the dose adjustments were done according to a clinical guideline, the patients were not blinded to the intervention (ie, the study was open label). The inclusion of a control group would have expanded the possibilities for interpretation of the results (eg, by providing flare rates for patients undergoing vs not undergoing dose reduction). The clinical remission criteria could be debated. BASDAI < 40 could be considered high, but it was only 1 of several criteria that needed to be fulfilled for a patient to be considered in clinical remission.
In summary, 99% of patients in clinical remission who tapered TNFi according to a clinical guideline flared before 2-year follow-up, but slightly more than half of the patients remained in remission on half dose. Most flares occurred approximately 12-16 weeks after tapering. ASAS-defined clinically important worsening was experienced by 41% of patients, and the most frequent symptoms were back and buttock pain and pain in peripheral joints and entheseal regions. Physician global evaluation of disease activity was an independent predictor of flare after tapering. The fact that more than half of patients with axSpA did not flare until tapered to one-third of standard dose or less encourages structured tapering regimens with dose step-ups at flares.
ACKNOWLEDGMENT
We acknowledge the contribution of secretaries, laboratory technicians, radiographers, nurses, and physicians in the implementation of the guideline. The work was presented at American College of Rheumatology’s 2021 annual meeting (abstract no. 0929, Arthritis Rheumatol, 2021; 73 [Suppl 10]) and at the European Alliance of Associations for Rheumatology’s 2022 annual meeting (abstract no. 654, Ann Rheum Dis 2022, 81 [Suppl 1] 396).
Footnotes
This quality control study was supported by Center for Rheumatology and Spine Diseases, Center for Head and Orthopaedics, University Hospital of Copenhagen, Rigshospitalet. MW was supported with a PhD fellowship grant (R131-A5381) from the Danish Rheumatism Association.
AH has received speaker fees from Eli Lilly. MB has received research support, consultancy fees, and/or speaker fees from Image Analysis Group, Esaote, AbbVie, Celgene, Eli Lilly, Janssen, Novartis, Pfizer, UCB, Novo, GSK, Takeda, Geurbet, Biogen, Radiobotics, and Chondrometrics. MLH has received grants from BMS, AbbVie, Roche, and Novartis; grants and personal fees from MSD, Biogen, and Pfizer; and personal fees from Eli Lilly, Orion, Celltrion, Samsung Bioepis, and Janssen Biologics BV. MLH chairs the steering committee of the Danish Rheumatology Quality Registry (DANBIO), which receives public funding from the hospital owners and funding from pharmaceutical companies; and cochairs EuroSpA, which generates real-world evidence of treatment of psoriatic arthritis and axial spondylorthritis based on secondary data and is partly funded by Novartis. MØ has received research support, consultancy fees, and/or speaker fees from AbbVie, BMS, Boehringer Ingelheim, Celgene, Eli Lilly, Hospira, Janssen, Merck, Novartis, Pfizer, Regeneron, Roche, Sandoz, Sanofi, and UCB. SNC has received speakers’ fees from BMS and GE and research support from Novartis. SJP has been an advisory board member for AbbVie and Novartis; received research support from AbbVie, MSD, and Novartis; and received speaker fees from MSD, Pfizer, AbbVie, Novartis, and UCB. The remaining authors declare no conflicts of interest relevant to this article.
- Accepted for publication October 5, 2023.
- Copyright © 2024 by the Journal of Rheumatology
