Early Improvements in Disease Activity Indices Predict Long-Term Clinical Remission Suggested by the Treat-to-Target Strategy in Patients with Ankylosing Spondylitis Receiving TNF-α Inhibitor Treatment

This study evaluated the possibility of clinical remission suggested by the treat-to-target strategy and identified predictors of clinical remission in 139 patients with ankylosing spondylitis (AS) receiving tumor necrosis factor-α inhibitors (TNFi). Clinical remission criteria selected were AS Disease Activity Score Inactive Disease (ASDAS-ID) and Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) < 2 with normal C-reactive protein (CRP) levels (BASDAI-CRP). The longitudinal relationship between clinical parameters and clinical remission was assessed using generalized estimating equations (GEEs). Responders to ASDAS-ID and BASDAI-CRP increased from 32.4% to 68.9% and from 39.9% to 75.2% at months 3 and 33, respectively. Responders to ASDAS-ID and BASDAI-CRP almost overlapped. In the univariable GEE model, age and 3-month improvement in BASDAI, ASDAS-CRP, physician and patient global assessments, and spinal pain predicted clinical remission achievement, while the presence of syndesmophytes predicted ASDAS-CRP achievement, and normalized CRP at 3 months was associated with BASDAI-CRP achievement. Multivariable GEE analysis revealed age (odds ratio (OR): 0.67; 95% confidence interval (CI), 0.49–0.93) and 3-month BASDAI improvement (OR: 1.70; CI, 1.19–2.41) as independent predictors of ASDAS-ID achievement and age (OR: 0.69; CI, 0.54–0.89), 3-month BASDAI improvement (OR: 2.00; CI, 1.45–2.76), and normalized CRP at 3 months (OR: 3.72; CI, 1.39–9.95) as independent predictors of BASDAI-CRP achievement.


Introduction
Over the past decade, tumor necrosis factor-α inhibitors (TNFi) have become the cornerstone for the treatment of ankylosing spondylitis (AS). These improve functional outcomes and decrease axial symptoms, peripheral arthritis, enthesitis, and extra-articular symptoms in patients with marginal benefit from non-steroidal anti-inflammatory drug (NSAID) therapy [1,2]. Furthermore, the Assessment of SpondyloArthritis International Society (ASAS)-European League Against Rheumatism (EULAR) [3] and American College of Rheumatology (ACR)/Spondylitis Association of America (SAA)/Spondyloarthritis Research and Treatment Network (SPARTAN) have recommended TNFi as the first biological agent in the treatment of AS [4,5].
As TNFi significantly improves clinical outcomes in patients with spondyloarthritis (SpA), treat-to-target (T2T) is an emerging management strategy for SpA [6][7][8]. The concept of T2T in SpA was developed based on the results of a systematic literature review and expert opinion by an international task force in 2012 and updated in 2017 [6,7]. According to this strategy, validated composite measures of disease activity, such as the AS Disease Activity Score (ASDAS) or Bath AS Disease Activity Index (BASDAI) with acute phase reactants (APRs), were recommended to be performed in order to measure clinical disease activity. The therapeutic target should be clinical remission, and low disease activity may be an alternative target. Clinical remission is mainly reflected by ASDAS-ID, low BASDAI with normal C-reactive protein (CRP) level, or ASAS partial remission (ASAS-PR).
For the measure of disease activity, ASDAS is a preferred method because it provides better disease discrimination than BASDAI and is more associated with various biomarkers of inflammation than BASDAI [6,7]. In case of BASDAI with APRs, BASDAI has some limitations in that its questions are only relevant to patient-reported outcomes and not specific for symptoms related to inflammation versus other processes [9]. Additionally, the definition of low BASDAI differs among studies due to the lack of a validated definition of clinical remission or low disease activity for BASDAI [10]. However, its ease of use, reproducibility, and sensitivity to changes make BASDAI a useful instrument for research purposes, and it has been used to assess the response to TNFi treatment in patients with AS [11]. Furthermore, APRs, which are objective measurements of inflammation, may complement BASDAI.
Many studies have evaluated the therapeutic effects of TNFi in patients with AS. Several studies that assessed predictors of clinical outcome in AS patients treated with TNFi used ASDAS-ID as the treatment target [12][13][14][15], and a few studies used low BASDAI with normal CRP as the target [13]. Although earlier studies identified only baseline clinical parameters as predictors for clinical remission [14,16,17], subsequent studies reported early clinical response as predictors of treatment outcome [12,13,15].
In this study, we aimed to evaluate the possibility of achieving clinical remission suggested by the T2T strategy, ASDAS-ID and low BASDAI with normal CRP, during long-term TNFi treatment and to investigate the predictors of clinical remission.

Patients
We performed a retrospective observational study on 139 patients with AS who were treated with TNFi for up to 33 months between January 2006 and December 2019. All patients were aged >19 years, fulfilled the 1984 modified New York criteria for AS [18], and had begun TNFi therapy according to the Korean National Health Insurance guidelines, which recommend TNFi for use in patients who have active disease with BASDAI of ≥40/100, despite concurrent treatment with two or more anti-rheumatic drugs, including NSAIDs, for at least 3 months. At 3 months and every 6 months thereafter, TNFi treatment continuation was determined based on BASDAI improvement (50% improvement in BASDAI (BASDAI 50 ) or absolute improvement of ≥2 units, compared with baseline), which was used as the definition of treatment response according to the Korean National Health Insurance guidelines. This study was approved by the Ethics Committee of the Kyungpook National University Hospital (KNUH 2020-12-030). Informed consent was waived due to the retrospective nature of the study.

Statistical Analysis
Descriptive statistics are expressed as mean ± standard deviation (SD). Continuous variables were compared using Student's t-test and categorical variables using the chisquare or Fisher's exact test. The paired t-test was used to compare outcome measures between visits. Bivariate correlations were determined using Spearman's correlation coefficient. Longitudinal relationships between clinical parameters and achievement of clinical remission were assessed using generalized estimating equations (GEEs), with ASDAS-ID and BASDAI-CRP as dependent variables and individual clinical parameters as independent variables [21]. If a parameter demonstrated a significant association with the outcome in the univariable GEE model (p < 0.05), it was included as a covariate in the multivariable model. An "exchangeable" correlation structure was most appropriate for our study based on the correlation coefficients among outcomes at different follow-up evaluations [19]. Quasi-likelihood under the independence model criterion (QIC) was used to estimate the model fitness, and lower QICs reflected better data fit of the model. All statistical tests were performed at a significance level of α = 0.05. Statistical analyses were performed using SPSS (ver. 21.0; IBM, Armonk, NY, USA).

Early Improvement in Disease Activity Indices as a Predictor of ASDAS-ID Achievement during TNFi Treatment
We evaluated the predictors of ASDAS-ID achievement during TNFi treatment by using the GEE model ( Table 2). The analysis using univariable GEE with baseline parameters, including age, disease duration, male sex, presence of HLA and syndesmophytes, and history of uveitis, demonstrated that older age (odds ratio (OR): 0.71, per 10 years increase; 95% confidence interval (CI), 0.54-0.92; p = 0.009) and presence of syndesmophytes (OR = 0.55; CI, 0.32-0.97; p = 0.038) were associated with a significantly lower possibility of achieving ASDAS-ID during TNFi treatment. Disease duration, presence of HLA-B27, and male sex did not influence the ASDAS-ID achievement. As TNFi induced significant improvements in disease activity and functional and metrological indices with the most prominent change at 3 months ( Figures S4 and S5), we conducted the univariable GEE models with 3-month improvements in these parameters to assess ASDAS-ID achievement. All 3-month improvements in disease activity indices, including ASDAS-CRP, BASDAI, PhGA, PtGA, and spinal pain, were associated with long-term ASDAS-ID achievement, whereas improvements in functional (BASFI) and metrological (BASMI 10 and chest expansion) indices and APRs were not. Since age and the presence of syndesmophytes were predictors of ASDAS-ID achievement among baseline variables, 3-month improvements in these parameters were analyzed after adjustment for age and presence of syndesmophytes on the GEE model.

Early Change in Disease Activity Indices as a Predictor of BASDAI-CRP Attainment during TNFi Treatment
In the univariable GEE analysis using baseline variables, age (OR = 0.76, per 10 years increase; CI, 0.60-0.96; p = 0.021) was a negative predictor of BASDAI-CRP achievement ( Table 3) Due to the fact that the proportion of normalized ESR or CRP at 3 months was similar (ESR, 85.8% (118/138); CRP, 87.0% (120/138)) and the 3-month improvement in ESR was significantly correlated with that in CRP (r = 0.712, p < 0.001), we performed multivariable GEE analysis for BASDAI-CRP achievement without the variable of normalized ESR at 3 months. On the multivariable GEE analysis, age (OR = 0.69, per 10 years increase; CI, 0.54-0.89); p = 0.004), 3-month BASDAI improvement (OR = 2.00; CI, 1.45-2.76; p < 0.001), and normalized CRP at 3 months (OR = 3.72; CI, 1.39-9.95; p = 0.009) were independent predictors of BASDAI-CRP achievement, and the QIC was 799.491. Table 3. Effect of improvement in disease activity indices at 3 months on the achievement of low Bath Ankylosing Spondylitis Disease Activity Index and normal C-reactive protein level (BASDAI-CRP) during the treatment with tumor necrosis factor-α inhibitor in patients with ankylosing spondylitis.

Sub-Analysis of Identified Predictors for Achievement of Clinical Remission during TNFi Treatment
As younger age was associated with a higher achievement rate of both clinical remission criteria, we analyzed the proportion of attainment of ASDAS-ID and BASDAI-CRP using the age of 40 years as cut-off; patients aged <40 years (n = 59) exhibited about two times the likelihood of achieving clinical remission than those aged ≥40 years (n = 80) (ASDAS-ID, OR = 2.05; CI, 1.16-3.62; p = 0.014 and BASDAI-CRP, OR = 1.99; CI, 1.17-3.37; p = 0.011). A significant difference in achievement rate based on the age of 40 years was observed at 9 months (p = 0.019), 15 (p = 0.005), and 27 (p = 0.037) in ASDAS-ID and at 3 months (p = 0.009) in BASDAI-CRP (Figure 2A,B).

Discussion
In this study, the proportion of those who achieved BASDAI-CRP was higher than that of patients who achieved ASDAS-ID, and patients who satisfied the ASDAS-ID and BASDAI-CRP almost overlapped. In the multivariable GEE analysis, age and 3-month BASDAI improvement were independent predictors of the achievement of both clinical remission criteria. On the other hand, normalized CRP at 3 months was an independent predictor of BASDAI-CRP achievement. With respect to age, patients aged < 40 years showed two-times higher rate of achievement of both clinical remission criteria. For a 3-month BASDAI improvement ≥ 3, the probability of ASDAS-ID attainment increased by 2.7 times and that of BASDAI-CRP increased by 2.2 times.
The concept of T2T strategies has been widely used to obtain optimal outcomes in chronic non-rheumatic diseases, such as hypertension and hyperlipidemia [24]. T2T strategies have also been incorporated in the management of rheumatic diseases, especially rheumatoid arthritis (RA) [8,25]. This treatment approach in RA is now a well-established practice with regular disease activity monitoring using validated composite measures of disease activity and remission criteria. The T2T approach in SpA was recommended by an international task force [6], but it has not yet been established because of the lack of robust direct evidence and no absolute consensus about which therapeutic target to use [5,8].
Although there is no widely accepted definition for clinical remission in the T2T strategy in AS, ASDAS-ID was suggested by the task force as it comprises the estimation of clinical aspects of the disease as well as APRs [6][7][8]. Low BASDAI with normal CRP level and ASAS-PR were alternatively suggested as clinical remission criteria [6,7]. We selected both ASDAI-CRP and BASDAI-CRP as clinical remission criteria because BASDAI incorporates additional clinical outcomes of interest, such as physical function; assesses enthesopathy; and has both eligibility criterion and outcome in clinical trials [9]. In addition, in clinical practice, the BASDAI measurement is an easier striker compared to the ASAS-PR assessment. Moreover, CRP, an objective parameter, may compensate for some of the shortcomings of BASDAI. Due to the unvalidated definition of low disease activity or clinical remission for BASDAI [10], different cut-off values have been used in observational [26], cohort [19], and controlled [13,27] studies as follows: BASDAI < 4 [26], BASDAI < 3 [27], and BASDAI < 2 [13,19]. In this study, we adopted the strictest cut-off value, BASDAI < 2, which was used in cohort [19] and randomized controlled [13] studies. We found patients who satisfied the ASDAS-ID, and BASDAI-CRP almost overlapped, but the achievement rate of BASDAI-CRP was higher than that of ASDAS-ID. This may be because both clinical remission criteria share common components, such as CRP and variables of back pain, duration of morning stiffness, and peripheral joint symptoms in BASDAI, whereas ASDAS-ID achievement needs PtGA improvement as another component. Furthermore, BASDAI < 2 as the cut-off value may not imply an inactive disease status [11].
Early studies demonstrated that younger age, male sex, HLA-B27 positivity, higher disease activity, increased CRP, and higher functional status were independent baseline predictors of the clinical response to TNFi treatment [14,16,17]. Subsequent studies showed that early response to TNFi predicted long-term clinical improvement [12,13,15], and the effect of baseline characteristics on the long-term clinical improvement was relatively small compared with the effect of early clinical response [12].
In our study, among baseline demographic factors, age was an independent predictor of achievement of both clinical remission criteria. Older patients usually had longer disease recovery periods, while younger patients tended to have shorter disease duration. Age was closely correlated with disease duration (r = 0.569, p < 0.001). These results suggest that because young age is related with early diagnosis, young patients have a great effect on TNFi treatment. However, disease duration did not predict clinical remission achievement in our study. Several studies also demonstrated that age was associated with TNFi response, whereas disease duration was not [14,16,28], which suggests that there may be other mechanisms in the relationship between age and TNFi response. One suggested underlying explanation is that young age is related with low percentage of TNFα-producing CD8+ T cells in AS [28]. Absence of syndesmophytes was also a positive predictor of ASDAS-ID achievement. Disease duration was not associated with the achievement of clinical remission criteria, which may be due to the long disease duration of the enrolled patients in our study. As the clinical improvements at 3 months were the most prominent with TNFi treatment in our study, we tried to demonstrate that early clinical improvements were associated with a higher possibility of sustained attainment of long-term clinical remission. In terms of achievements of both clinical remission criteria, the 3-month BASDAI improvement was the strongest predictor. Higher 3-month BASDAI improvement, especially BASDAI improvement of ≥3, was an independent predictor of achievement of long-term clinical remission criteria in both ASDAS-ID and BASDAI-CRP. On the other hand, the 3-month ASDAS-CRP improvement was not a strong predictor of achievement of clinical remission criteria on the univariable GEE model and was not an independent predictor on multiple GEE model. Considering that ASDAS-CRP shares components of BASDAI, multivariable GEE model without the 3-month BASDAI improvement for ASDAS-ID achievement was also performed. However, even in this case, the 3-month PhGA, not the 3-month ASDAS-CRP improvement, was an independent predictor of ASDAS-ID achievement (data not shown). Normalized CRP at 3 months was an independent predictor of BASDAI-CRP, not ASDAS-ID, achievement, which may be related to the fact that BASDAI-CRP achievement needs both CRP normalization and low BASDAI, while ASDAS-ID achievement demands not only CRP improvement but also improvements in other four variables. With respect to the 3-month improvements, PhGA, PtGA, and spinal pain were associated with the attainment of ASDAS-ID and BASDAI-CRP in the univariable GEE model but not in the multivariable GEE model. It is possible that the interaction of these three factors resulted in a loss of statistical significance. The 3-month improvements in functional (BASFI) and metrological (BASMI 10 and chest expansion) indices were not associated with long-term clinical remission achievement, which may be related with irreversible structural changes due to longer disease duration.
Our study has some limitations. First, because we selected BASDAS < 2 as the cut-off for clinical remission for BASDAI, analysis using different cut-offs may result in divergence of results in identifying the predictors of attainment of clinical remission. Second, because this was a retrospective and observational study, bias may be present. For example, if a description of the reason for elevated CRP was lacking, whether it was caused by disease activity or by accompanying infection, it had to be inferred by using other disease activity indices. Further studies are required to determine whether BASDAI-CRP achievement can be a clinical remission criterion in T2T strategy after the validation of the cut-off value of low disease activity in BASDAI during long-term TNFi treatment in AS.
In conclusion, we found that the proportion of patients who achieved BASDAI-CRP was higher than those who achieved ASDAS-ID, and patients who satisfied the ASDAS-ID and BASDAI-CRP almost overlapped. Age and 3-month BASDAI improvement were independent predictors of achievement of both long-term clinical remission criteria, and the normalized CRP at 3 months was an independent predictor of BASDAI-CRP achievement.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/jcm10184279/s1, Figure S1: Response to TNF-α inhibitor (TNFi) treatment as measured by the ASDAS inactive disease (ASDAS-ID) and low BASDAI with normal CRP levels (BASDAI-CRP). Index with normal C-reactive protein level. * p < 0.05, ** p < 0.01, Figure S3: Difference in metrological indices, (A and B) BASMI 10 , and (C and D) chest expansion depending on satisfaction of ASDAS-ID and BASDAI-CRP. ASDAS-ID, Ankylosing Spondylitis Disease Activity Score Inactive Disease; BASDAI-CRP, low Bath Ankylosing Spondylitis Metrology Index with normal C-reactive protein level. * p < 0.05, ** p < 0.01, Figure S4: Improvements in (A-E) disease activity and (F) functional outcomes over 33 months after TNF-α inhibitor treatment in patients with AS. The p-values represent differences in data between visits using the paired t-test. ASDAS-CRP, Ankylosing Spondylitis Disease Activity Score based on C-reactive protein; BASDAI, Bath Ankylosing Spondylitis Metrology Index; PhGA, physician global assessment; PtGA, patient global assessment; BASFI, Bath Ankylosing Spondylitis Functional Index. * p < 0.05, ** p < 0.01, Figure S5: Improvements in (A,B) metrological indices and (C,D) acute phase reactants over 33 months with TNF-α inhibitor treatment in patients with AS. p-values represent differences in data between visits using the paired t-test. BASMI, Bath Ankylosing Spondylitis Metrological Index; ChE, chest expansion; ESR, erythrocyte sedimentation rate; CRP, C-reactive protein. * p < 0.05, ** p < 0.01, Table S1: Response to treatment with tumor necrosis factor-α inhibitor as measured by the criteria of the Ankylosing Spondylitis Disease Activity Score (ASDAS), low Bath Ankylosing Spondylitis Disease Activity Index with normal C-reactive protein level (BASDAI-CRP), and at least 50% improvement in the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI 50 ).  Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the Kyungpook National University Hospital (KNUH 2020-12-030).
Informed Consent Statement: Informed consent was waived due to the retrospective nature of the study.
Data Availability Statement: All data generated or analyzed during this study are included in the study.